═══ 1. The Watcom C/C++ Compile and Link Utility ═══ The Watcom C/C++ Compile and Link Utility is designed for generating applications, simply and quickly, using a single command line. On the command line, you can list source file names as well as object file names. Source files are either compiled or assembled based on file extension; object files and libraries are simply included in the link phase. Options can be passed on to both the compiler and linker. ═══ 1.1. WCL/WCL386 Command Line Format ═══ The format of the command line is: WCL [files] [options] WCL386 [files] [options] The square brackets [ ] denote items which are optional. WCL is the name of the Watcom Compile and Link utility that invokes the 16-bit compiler. WCL386 is the name of the Watcom Compile and Link utility that invokes the 32-bit compiler. The files and options may be specified in any order. The Watcom Compile and Link utility uses the extension of the file name to determine if it is a source file, an object file, or a library file. Files with extensions of "OBJ" and "LIB" are assumed to be object files and library files respectively. Files with extensions of "ASM" are assumed to be assembler source files and will be assembled by the Watcom Assembler. Files with any other extension, including none at all, are assumed to be C/C++ source files and will be compiled. Pattern matching characters ("*" and "?") may be used in the file specifications. If no file extension is specified for a file name then the Watcom Compile and Link utility will check for a file with one of the following extensions. Order Name.Ext Assumed to be ----- -------- --------------- 1. file.ASM Assembler source code 2. file.CXX C++ source code 3. file.CPP C++ source code 4. file.CC C++ source code 5. file.C C source code It checks for each file in the order listed. By default, the Watcom Assembler will be selected to compile files with the extension "ASM". By default, the Watcom C++ compiler will be selected to compile files with any of the extensions "CXX", "CPP" or "CC". By default, the Watcom C compiler will be selected to compile a file with a "C" extension. The default selection of compiler can be overridden by the "cc" and "cc++" options, described below. Options are prefixed with a slash (/) or a dash (-) and may be specified in any order. Options can include any of the Watcom C/C++ compiler options plus some additional options specific to the Watcom Compile and Link utility. A summary of options is displayed on the screen by simply entering the "WCL" or "WCL386" command with no arguments. ═══ 1.2. Watcom Compile and Link Options Summary ═══ c compile the files only, do not link them cc treat source files as C code cc++ treat source files as C++ code y ignore the WCL/WCL386 environment variable 0 (16-bit only) 8088 and 8086 instructions (default for 16-bit) 1 (16-bit only) 188 and 186 instructions 2 (16-bit only) 286 instructions 3 (16-bit only) 386 instructions 4 (16-bit only) 486 instructions 5 (16-bit only) Pentium instructions 6 (16-bit only) Pentium Pro instructions 3r (32-bit only) generate 386 instructions based on 386 instruction timings and use register-based argument passing conventions 3s (32-bit only) generate 386 instructions based on 386 instruction timings and use stack-based argument passing conventions 4r (32-bit only) generate 386 instructions based on 486 instruction timings and use register-based argument passing conventions 4s (32-bit only) generate 386 instructions based on 486 instruction timings and use stack-based argument passing conventions 5r (32-bit only) generate 386 instructions based on Intel Pentium instruction timings and use register-based argument passing conventions (default for 32-bit) 5s (32-bit only) generate 386 instructions based on Intel Pentium instruction timings and use stack-based argument passing conventions 6r (32-bit only) generate 386 instructions based on Intel Pentium Pro instruction timings and use register-based argument passing conventions 6s (32-bit only) generate 386 instructions based on Intel Pentium Pro instruction timings and use stack-based argument passing conventions bc (C++ only) build target is a console application bd build target is a Dynamic Link Library (DLL) bg (C++ only) build target is a GUI application bm build target is a multi-thread environment br build target uses DLL version of C/C++ run-time libraries bt[=] build target for operating system bw build target uses default windowing support d0 (C++ only) no debugging information d1 line number debugging information d1+ (C only) line number debugging information plus typing information for global symbols and local structs and arrays d2 full symbolic debugging information d2i (C++ only) d2 and debug inlines; emit inlines as external out-of-line functions d2s (C++ only) d2 and debug inlines; emit inlines as static out-of-line functions d2t (C++ only) full symbolic debugging information, without type names d3 full symbolic debugging with unreferenced type names ,* d3i (C++ only) d3 plus debug inlines; emit inlines as external out-of-line functions d3s (C++ only) d3 plus debug inlines; emit inlines as static out-of-line functions d[=text] preprocessor #define name [text] d+ allow extended -d macro definitions db generate browsing information e set error limit number (default is 20) ee call epilogue hook routine ef use full path names in error messages ei force enum base type to use at least an int em force enum base type to use minimum en emit routine name before prologue ep[] call prologue hook routine with number of stack bytes available eq do not display error messages (they are still written to a file) er (C++ only) do not recover from undefined symbol errors et Pentium profiling ew (C++ only) generate less verbose messages ez (32-bit only) generate Phar Lap Easy OMF-386 object file fc= (C++ only) specify file of command lines to be batch processed fh[q][=] use precompiled headers fhd store debug info for pre-compiled header once (DWARF only) fhr (C++ only) force compiler to read pre-compiled header fhw (C++ only) force compiler to write pre-compiled header fhwe (C++ only) don't include pre-compiled header warnings when "we" is used fi= force file_name to be included fo= set object or preprocessor output file specification fpc generate calls to floating-point library fpi (16-bit only) generate in-line 80x87 instructions with emulation (default) (32-bit only) generate in-line 387 instructions with emulation (default) fpi87 (16-bit only) generate in-line 80x87 instructions (32-bit only) generate in-line 387 instructions fp2 generate in-line 80x87 instructions fp3 generate in-line 387 instructions fp5 generate in-line 80x87 instructions optimized for Pentium processor fp6 generate in-line 80x87 instructions optimized for Pentium Pro processor fpd enable generation of Pentium FDIV bug check code fpr generate 8087 code compatible with older versions of compiler fr= set error file specification ft (C++ only) try truncated (8.3) header file specification fx (C++ only) do not try truncated (8.3) header file specification g= set code group name h{w,d,c} set debug output format (Watcom, Dwarf, Codeview) i= add directory to list of include directories j change char default from unsigned to signed k (C++ only) continue processing files (ignore errors) m{f,s,m,c,l,h} memory model - mf=flat ms=small mm=medium mc=compact ml=large mh=huge (default is "ms" for 16-bit and Netware, "mf" for 32-bit) nc= set name of the code class nd= set name of the "data" segment nm= set module name different from filename nt= set name of the "text" segment o{a,b,c,d,e,f,f+,h,i,i+,k,l,l+,m,n,o,p,r,s,t,u,x,z} control optimization p{e,l,c,w=} preprocess file only, sending output to standard output; "c" include comments; "e" encrypt identifiers (C++ only); "l" include #line directives; w= wrap output lines at columns (zero means no wrap) r save/restore segment registers ri return chars and shorts as ints s remove stack overflow checks sg generate calls to grow the stack st touch stack through SS first t= (C++ only) set tab stop multiplier u preprocessor #undef name v output function declarations to .def file (with typedef names) vc... (C++ only) VC++ compatibility options w set warning level number (default is w1) wcd= warning control: disable warning message wce= warning control: enable warning message we treat all warnings as errors wo (C only) (16-bit only) warn about problems with overlaid code wx set warning level to maximum setting xd (C++ only) disable exception handling (default) xdt (C++ only) disable exception handling (same as "xd") xds (C++ only) disable exception handling (table-driven destructors) xr (C++ only) enable RTTI xs (C++ only) enable exception handling xst (C++ only) enable exception handling (direct calls for destruction) xss (C++ only) enable exception handling (table-driven destructors) z{a,e} disable/enable language extensions (default is ze) zc place literal strings in code segment zd{f,p} allow DS register to "float" or "peg" it to DGROUP (default is zdp) zdl (32-bit only) load DS register directly from DGROUP zf{f,p} allow FS register to be used (default for all but flat memory model) or not be used (default for flat memory model) zg output function declarations to .def (without typedef names) zg{f,p} allow GS register to be used or not used zk0 double-byte char support for Kanji zk0u translate Kanji double-byte characters to UNICODE zk1 double-byte char support for Chinese/Taiwanese zk2 double-byte char support for Korean zkl double-byte char support if current code page has lead bytes zku= load UNICODE translate table for specified code page zl suppress generation of library file names and references in object file zld suppress generation of file dependency information in object file zm place each function in separate segment (near functions not allowed) zmf place each function in separate segment (near functions allowed) zp[{1,2,4,8,16}] set minimal structure packing (member alignment) (default is zp1) zpw output warning when padding is added in a struct/class zq operate quietly zs syntax check only zt set data threshold (default is zt32767) zu do not assume that SS contains segment of DGROUP zv (C++ only) enable virtual function removal optimization zw Microsoft Windows prologue/epilogue code sequences zW (16-bit only) Microsoft Windows optimized prologue/epilogue code sequences zWs (16-bit only) Microsoft Windows smart callback sequences zz remove "@size" from __stdcall function names (10.0 compatible) See the Watcom C/C++ User's Guide for a full description of compiler options. k set stack size fd[=] keep directive file and, optionally, rename it (default name is "__WCL__.LNK"). fe= name executable file fm[=] generate map file and name it (optional) lp (16-bit only) create an OS/2 protected-mode program lr (16-bit only) create a DOS real-mode program l= link a program for the specified system. Among the supported systems are: 286 16-bit DOS executables (synonym for "DOS") under DOS and NT hosted platforms; 16-bit OS/2 executables (synonym for "OS2") under 32-bit OS/2 hosted OS/2 session. 386 32-bit DOS executables (synonym for "DOS4G") under DOS; 32-bit NT character-mode executables (synonym for "NT") under Windows NT; 32-bit OS/2 executables (synonym for "OS2V2") under 32-bit OS/2 hosted OS/2 session. ADS 32-bit AutoCAD ADS executables COM 16-bit DOS "COM" files DOS 16-bit DOS executables DOS4G 32-bit Tenberry Software DOS Extender executables DOS4GNZ 32-bit Tenberry Software DOS Extender non-zero base executables EADI 32-bit AutoCAD ADI executables (emulation) FADI 32-bit AutoCAD ADI executables (floating-point) NETWARE 32-bit Novell NetWare 386 NLMs NOVELL 32-bit Novell NetWare 386 NLMs (synonym for NETWARE) NT 32-bit Windows NT character-mode executables NT_DLL 32-bit Windows NT DLLs NT_WIN 32-bit Windows NT windowed executables OS2 16-bit OS/2 V1.x executables OS2V2 32-bit OS/2 executables OS2V2_PM 32-bit OS/2 PM executables PHARLAP 32-bit PharLap DOS Extender executables QNX 16-bit QNX executables QNX386 32-bit QNX executables TNT 32-bit Phar Lap TNT DOS-style executable WIN386 32-bit extended Windows 3.x executables/DLLs WIN95 32-bit Windows 95 executables/DLLs WINDOWS 16-bit Windows executables WINDOWS_DLL 16-bit Windows Dynamic Link Libraries X32R 32-bit FlashTek (register calling convention) executables X32RV 32-bit FlashTek Virtual Memory (register calling convention) executables X32S 32-bit FlashTek (stack calling convention) executables X32SV 32-bit FlashTek Virtual Memory (stack calling convention) executables These names are among the systems identified in the Watcom Linker initialization file, "WLSYSTEM.LNK". The Watcom Linker "SYSTEM" directives, found in this file, are used to specify default link options for particular (operating) systems. Users can augment the Watcom Linker initialization file with their own system definitions and these may be specified as an argument to the "l=" option. The "system_name" specified in the "l=" option is used to create a "SYSTEM system_name" Watcom Linker directive when linking the application. x make names case sensitive @ include additional directive file "linker directives" allows use of any linker directive ═══ 1.3. WCL/WCL386 Environment Variables ═══ The WCL environment variable can be used to specify commonly used WCL options. The WCL386 environment variable can be used to specify commonly used WCL386 options. These options are processed before options specified on the command line. Example: C>set wcl=/d1 /ot C>set wcl386=/d1 /ot The above example defines the default options to be "d1" (include line number debugging information in the object file), and "ot" (favour time optimizations over size optimizations). Whenever you wish to specify an option that requires the use of an "=" character, you can use the "#" character in its place. This is required by the syntax of the "SET" command. Once the appropriate environment variable has been defined, those options listed become the default each time the WCL or WCL386 command is used. The WCL environment variable is used by WCL only. The WCL386 environment variable is used by WCL386 only. Both WCL and WCL386 pass the relevant options to the Watcom C/C++ compiler and linker. This environment variable is not examined by the Watcom C/C++ compiler or the linker when invoked directly. Hint: If you are running DOS and you use the same WCL or WCL386 options all the time, you may find it handy to place the "SET WCL" or "SET WCL386" command in your DOS system initialization file, AUTOEXEC.BAT. If you are running OS/2 and you use the same WCL or WCL386 options all the time, you may find it handy to place the "SET WCL" or "SET WCL386" command in your OS/2 system initialization file, CONFIG.SYS. ═══ 1.4. WCL/WCL386 Command Line Examples ═══ For most small applications, the WCL or WCL386 command will suffice. We have only scratched the surface in describing the capabilities of the WCL and WCL386 commands. The following examples describe the WCL and WCL386 commands in more detail. Suppose that your application is contained in three files called APDEMO.C, APUTILS.C, and APDATA.C. We can compile and link all three files with one command. Example 1: C>wcl /d2 apdemo.c aputils.c apdata.c C>wcl386 /d2 apdemo.c aputils.c apdata.c The executable program will be stored in APDEMO.EXE since APDEMO appeared first in the list. Each of the three files is compiled with the "d2" debug option. Debugging information is included in the executable file. We can issue a simpler command if the current directory contains only our three C/C++ source files. Example 2: C>wcl /d2 *.c C>wcl386 /d2 *.c WCL or WCL386 will locate all files with the ".c" filename extension and compile each of them. The name of the executable file will depend on which of the C/C++ source files is found first. Since this is a somewhat haphazard approach to naming the executable file, WCL and WCL386 have an option, "fe", which will allow you to specify the name to be used. Example 3: C>wcl /d2 /fe=apdemo *.c C>wcl386 /d2 /fe=apdemo *.c By using the "fe" option, the executable file will always be called APDEMO.EXE regardless of the order of the C/C++ source files in the directory. If the directory contains other C/C++ source files which are not part of the application then other tricks may be used to identify a subset of the files to be compiled and linked. Example 4: C>wcl /d2 /fe=apdemo ap*.c C>wcl386 /d2 /fe=apdemo ap*.c Here we compile only those C/C++ source files that begin with the letters "ap". In our examples, we have recompiled all the source files each time. In general, we will only compile one of them and include the object code for the others. Example 5: C>wcl /d2 /fe=apdemo aputils.c ap*.obj C>wcl386 /d2 /fe=apdemo aputils.c ap*.obj The source file APUTILS.C is recompiled and APDEMO.OBJ and APDATA.OBJ are included when linking the application. The ".obj" filename extension indicates that this file need not be compiled. Example 6: C>wcl /fe=demo *.c utility.obj C>wcl386 /fe=demo *.c utility.obj All of the C/C++ source files in the current directory are compiled and then linked with UTILITY.OBJ to generate DEMO.EXE. Example 7: C>set wcl=/mm /d1 /ox /k4096 C>wcl /fe=grdemo gr*.c graph.lib /fd=grdemo C>set wcl386=/d1 /ox /k4096 C>wcl386 /fe=grdemo gr*.c graph.lib /fd=grdemo All C/C++ source files beginning with the letters "gr" are compiled and then linked with GRAPH.LIB to generate GRDEMO.EXE which uses a 4K stack. The temporary linker directive file that is created by WCL or WCL386 will be kept and renamed to GRDEMO.LNK. Example 8: C>set libos2=c:\watcom\lib286\os2;c:\os2 C>set lib=c:\watcom\lib286\dos;c:\watcom\lib286 C>set wcl=/mm /lp C>wcl grdemo1 \watcom\lib286\os2\graphp.obj phapi.lib The file GRDEMO1 is compiled for the medium memory model and then linked with GRAPHP.OBJ and PHAPI.LIB to generate GRDEMO1.EXE which is to be used with Phar Lap's 286 DOS Extender. The "lp" option indicates that an OS/2 format executable is to be created. The file GRAPHP.OBJ in the directory "\WATCOM\LIB286\OS2" contains special initialization code for Phar Lap's 286 DOS Extender. The file PHAPI.LIB is part of the Phar Lap 286 DOS Extender package. The LIBOS2 environment variable must include the location of the OS/2 libraries and the LIB environment variable must include the location of the DOS libraries (in order to locate GRAPH.LIB). The LIBOS2 environment variable must also include the location of the OS/2 file DOSCALLS.LIB which is usually "C:\OS2". For more complex applications, you should use the "Make" utility. ═══ 2. The Watcom Assembler ═══ This chapter describes the Watcom Assembler. It takes as input an assembler source file (a file with extension ".asm") and produces, as output, an object file. The Watcom Assembler command line syntax is the following. WASM [options] [d:][path]filename[.ext] [options] [@env_var] The square brackets [ ] denote items which are optional. WASM is the name of the Watcom Assembler. d: is an optional drive specification such as "A:", "B:", etc. If not specified, the default drive is assumed. path is an optional path specification such as "\PROGRAMS\ASM\". If not specified, the current directory is assumed. filename is the file name of the assembler source file to be assembled. ext is the file extension of the assembler source file to be assembled. If omitted, a file extension of ".asm" is assumed. If the period "." is specified but not the extension, the file is assumed to have no file extension. options is a list of valid options, each preceded by a slash ("/") or a dash ("-"). Options may be specified in any order. The options supported by the Watcom Assembler are: {0,1,2,3,4,5}{p}{r,s} 0 same as ".8086" 1 same as ".186" 2{p} same as ".286" or ".286p" 3{p} same as ".386" or ".386p" (also defines "__386__" and changes the default USE attribute of segments from "USE16" to "USE32") 4{p} same as ".486" or ".486p" (also defines "__386__" and changes the default USE attribute of segments from "USE16" to "USE32") 5{p} same as ".586" or ".586p" (also defines "__386__" and changes the default USE attribute of segments from "USE16" to "USE32") p protect mode add r defines "__REGISTER__" add s defines "__STACK__" Example: /2 /3p /4pr /5p bt= defines "____" and checks the "_INCLUDE" environment variable for include files c do not output OMF COMENT records that allow WDISASM to figure out when data bytes have been placed in a code segment d[=text] define text macro d1 line number debugging support e stop reading assembler source file at END directive. Normally, anything following the END directive will cause an error. e set error limit number fe= set error file name fo= set object file name fi= force to be included fpc same as ".no87" fpi inline 80x87 instructions with emulation fpi87 inline 80x87 instructions fp0 same as ".8087" fp2 same as ".287" or ".287p" fp3 same as ".387" or ".387p" fp5 same as ".587" or ".587p" i= add directory to list of include directories j or s force signed types to be used for signed values m{t,s,m,c,l,h,f} memory model: (Tiny, Small, Medium, Compact, Large, Huge, Flat) -mt Same as ".model tiny" -ms Same as ".model small" -mm Same as ".model medium" -mc Same as ".model compact" -ml Same as ".model large" -mh Same as ".model huge" -mf Same as ".model flat" Each of the model directives also defines "____" (e.g., ".model small" defines "__SMALL__"). They also affect whether something like "foo proc" is considered a "far" or "near" procedure. nd= set data segment name nm= set module name nt= set name of text segment o allow C form of octal constants zq or q operate quietly ? or h print this message w set warning level number we treat all warnings as errors ═══ 2.1. Assembly Directives and Opcodes ═══ It is not the intention of this chapter to describe assembly-language programming in any detail. You should consult a book that deals with this topic. However, we present an alphabetically ordered list of the directives, opcodes and register names that are recognized by the assembler. .186 .286 .286c .286p .287 .386 .386p .387 .486 .486p .586 .586p .8086 .8087 aaa aad aam aas abs adc add addr ah al alias align .alpha and arpl assume at ax basic bh bl bound bp .break bsf bsr bswap bt btc btr bts bx byte c call callf casemap catstr cbw cdq ch cl clc cld cli clts cmc cmp cmps cmpsb cmpsd cmpsw cmpxchg cmpxchg8b .code comm comment common compact .const .continue cpuid cr0 cr2 cr3 cr4 .cref cs cwd cwde cx daa das .data .data? db dd dec df dh di div dl .dosseg dp dq dr0 dr1 dr2 dr3 dr6 dr7 ds dt dup dw dword dx eax ebp ebx echo ecx edi edx else elseif end endif endp ends .endw enter eq equ equ2 .err .errb .errdef .errdif .errdifi .erre .erridn .erridni .errnb .errndef .errnz error es esi esp even .exit export extern externdef extrn f2xm1 fabs fadd faddp far .fardata .fardata? farstack fbld fbstp fchs fclex fcom fcomp fcompp fcos fdecstp fdisi fdiv fdivp fdivr fdivrp feni ffree fiadd ficom ficomp fidiv fidivr fild fimul fincstp finit fist fistp fisub fisubr flat fld fld1 fldcw fldenv fldenvd fldenvw fldl2e fldl2t fldlg2 fldln2 fldpi fldz fmul fmulp fnclex fndisi fneni fninit fnop fnrstor fnrstord fnrstorw fnsave fnsaved fnsavew fnstcw fnstenv fnstenvd fnstenvw fnstsw for forc fortran fpatan fprem fprem1 fptan frndint frstor frstord frstorw fs fsave fsaved fsavew fscale fsetpm fsin fsincos fsqrt fst fstcw fstenv fstenvd fstenvw fstp fstsw fsub fsubp fsubr fsubrp ftst fucom fucomp fucompp fwait fword fxam fxch fxtract fyl2x fyl2xp1 ge global group gs gt high highword hlt huge idiv if if1 if2 ifb ifdef ifdif ifdifi ife ifidn ifidni ifnb ifndef ignore imul in inc include includelib ins insb insd insw int into invd invlpg invoke iret iretd irp ja jae jb jbe jc jcxz je jecxz jg jge jl jle jmp jmpf jna jnae jnb jnbe jnc jne jng jnge jnl jnle jno jnp jns jnz jo jp jpe jpo js jz label lahf lar large lds le lea leave length lengthof les .lfcond lfs lgdt lgs lidt .list .listall .listif .listmacro .listmacroall lldt lmsw local lock lods lodsb lodsd lodsw loop loope loopne loopnz loopz low lowword lroffset lsl lss lt ltr macro mask medium memory mod .model mov movs movsb movsd movsw movsx movzx mul name ne near nearstack neg no87 .nocref .nolist nop not nothing offset opattr option or org os_dos os_os2 out outs outsb outsd outsw para pascal pop popa popad popcontext popf popfd private proc proto ptr public purge push pusha pushad pushcontext pushf pushfd pword qword .radix rcl rcr rdmsr rdtsc readonly record rep repe .repeat repne repnz repz ret retf retn rol ror rsm sahf sal .sall sar sbb sbyte scas scasb scasd scasw sdword seg segment .seq seta setae setb setbe setc sete setg setge setl setle setna setnae setnb setnbe setnc setne setng setnge setnl setnle setno setnp setns setnz seto setp setpe setpo sets setz .sfcond sgdt shl shld short shr shrd si sidt size sizeof sldt small smsw sp ss st .stack .startup stc std stdcall sti stos stosb stosd stosw str struc struct sub sword syscall tbyte test textequ .tfcond this tiny tr3 tr4 tr5 tr6 tr7 typedef union .until use16 use32 uses vararg verr verw wait watcom_c wbinvd .while width word wrmsr xadd xchg .xcref xlat xlatb .xlist xor ═══ 2.2. Unsupported Directives ═══ Other assemblers support directives that this assembler does not. The following is a list of directives that are ignored by the Watcom Assembler (use of these directives results in a warning message). .alpha .cref .lfcond .list .listall .listif .listmacro .listmacroall .nocref .nolist page .sall .seq .sfcond subtitle subttl .tfcond title .xcref .xlist The following is a list of directives that are flagged by the Watcom Assembler (use of these directives results in an error message). addr .break casemap catstr .continue echo endmacro .endw .exit high highword invoke low lowword lroffset mask opattr option popcontext proto purge pushcontext .radix record .repeat .startup this typedef union .until .while width ═══ 2.3. Watcom Assembler Diagnostic Messages ═══ 1 Size doesn't match with previous definition 2 Invalid instruction with current CPU setting 3 LOCK prefix is not allowed on this instruction 4 REP prefix is not allowed on this instruction 5 Invalid memory pointer 6 Cannot use 386 addressing mode with current CPU setting 7 Too many base registers 8 Invalid index register 9 Scale factor must be 1, 2, 4 or 8 10 invalid addressing mode with current CPU setting 11 ESP cannot be used as index 12 Too many base/index registers 13 Memory offset cannot reference to more than one label 14 Offset must be relocatable 15 Memory offset expected 16 Invalid indirect memory operand 17 Cannot mix 16 and 32-bit registers 18 CPU type already set 19 Unknown directive 20 Expecting comma 21 Expecting number 22 Invalid label definition 23 Invalid use of SHORT, NEAR, FAR operator 24 No memory 25 Cannot use 386 segment register with current CPU setting 26 POP CS is not allowed 27 Cannot use 386 register with current CPU setting 28 Only MOV can use special register 29 Cannot use TR3, TR4, TR5 in current CPU setting 30 Cannot use SHORT with CALL 31 Only SHORT displacement is allowed 32 Syntax error 33 Prefix must be followed by an instruction 34 No size given before 'PTR' operator 35 Invalid IMUL format 36 Invalid SHLD/SHRD format 37 Too many commas 38 Syntax error: Unexpected colon 39 Operands must be the same size 40 Invalid instruction operands 41 Immediate constant too large 42 Can not use short or near modifiers with this instruction 43 Jump out of range 44 Displacement cannot be larger than 32k 45 Initializer value too large 46 Symbol already defined 47 Immediate data too large 48 Immediate data out of range 49 Can not transfer control to stack symbol 50 Offset cannot be smaller than WORD size 51 Can not take offset of stack symbol 52 Can not take segment of stack symbol 53 Segment too large 54 Offset cannot be larger than 32k 55 Operand 2 too big 56 Operand 1 too small 57 Too many arithmetic operators 58 Too many open square brackets 59 Too many close square brackets 60 Too many open brackets 61 Too many close brackets 62 Invalid number digit 63 Assembler Code is too long 64 Brackets are not balanced 65 Operator is expected 66 Operand is expected 67 Too many tokens in a line 68 Bracket is expected 69 Illegal use of register 70 Illegal use of label 71 Invalid operand in addition 72 Invalid operand in subtraction 73 One operand must be constant 74 Constant operand is expected 75 A constant operand is expected in addition 76 A constant operand is expected in subtraction 77 A constant operand is expected in multiplication 78 A constant operand is expected in division 79 A constant operand is expected after a positive sign 80 A constant operand is expected after a negative sign 81 Label is not defined 82 More than one override 83 Label is expected 84 Only segment or group label is allowed 85 Only register or label is expected in override 86 Unexpected end of file 87 Label is too long 88 This feature has not been implemented yet 89 Internal Error #1 90 Can not take offset of group 91 Can not take offset of segment 92 Invalid character found 93 Invalid operand size for instruction 94 This instruction is not supported 95 size not specified -- BYTE PTR is assumed 96 size not specified -- WORD PTR is assumed 97 size not specified -- DWORD PTR is assumed 500 Segment parameter is defined already 501 Model parameter is defined already 502 Syntax error in segment definition 503 'AT' is not supported in segment definition 504 Segment definition is changed 505 Lname is too long 506 Block nesting error 507 Ends a segment which is not opened 508 Segment option is undefined 509 Model option is undefined 510 No segment is currently opened 511 Lname is used already 512 Segment is not defined 513 Public is not defined 514 Colon is expected 515 A token is expected after colon 516 Invalid qualified type 517 Qualified type is expected 518 External definition different from previous one 519 Memory model is not found in .MODEL 520 Cannot open include file 521 Name is used already 522 Library name is missing 523 Segment name is missing 524 Group name is missing 525 Data emitted with no segment 526 Seglocation is expected 527 Invalid register 528 Cannot address with assumed register 529 Invalid start address 530 Label is already defined 531 Token is too long 532 The line is too long after expansion 533 A label is expected after colon 534 Must be associated with code 535 Procedure must have a name 536 Procedure is alreadly defined 537 Language type must be specified 538 End of procedure is not found 539 Local variable must immediately follow PROC or MACRO statement 540 Extra character found 541 Cannot nest procedures 542 No procedure is currently defined 543 Procedure name does not match 544 Vararg requires C calling convention 545 Model declared already 546 Model is not declared 547 Backquote expected 548 COMMENT delimiter expected 549 End directive required at end of file 550 Nesting level too deep 551 Symbol not defined 552 Insert Stupid warning #1 here 553 Insert Stupid warning #2 here 554 Spaces not allowed in command line options 555 Error: 556 Source File 557 No filename specified. 558 Out of Memory 559 Cannot Open File - 560 Cannot Close File - 561 Cannot Get Start of Source File - 562 Cannot Set to Start of Source File - 563 Command Line Contains More Than 1 File To Assemble 564 include path %s. 565 Unknown option %s. Use /? for list of options. 566 read more command line from %s. 567 Internal error in %s(%u) 568 OBJECT WRITE ERROR !! 569 NO LOR PHARLAP !! 570 Parameter Required 571 Expecting closing square bracket 572 Expecting file name 573 Floating point instruction not allowed with /fpc 574 Too many errors 575 Build target not recognised 576 Public constants should be numeric 0 written 577 Expecting symbol 578 Do not mix simplified and full segment definitions 579 Parms passed in multiple registers must be accessed separately, use %s 580 Ten byte variables not supported in register calling convention 581 Parameter type not recognised 582 forced error: 583 forced error: Value not equal to 0 : %d 584 forced error: Value equal to 0: %d 585 forced error: symbol defined: %s 586 forced error: symbol not defined: %s 587 forced error: string blank : <%s> 588 forced error: string not blank : <%s> 589 forced error: strings not equal : <%s> : <%s> 590 forced error: strings equal : <%s> : <%s> 591 included by file %s(%d) 592 macro called from file %s(%d) 593 Symbol %s not defined 594 Extending jump 595 Ignoring inapplicable directive 596 Unknown symbol class '%s' 597 Symbol class for '%s' already established 598 number must be a power of 2 599 alignment request greater than segment alignment 600 '%s' is already defined 601 %u unclosed conditional directive(s) detected ═══ 3. The Watcom Library Manager ═══ The Watcom Library Manager can be used to create and update object library files. It takes as input an object file or a library file and creates or updates a library file. For OS/2, Win16 and Win32 applications, it can also create import libraries from Dynamic Link Libraries. An object library is essentially a collection of object files. These object files generally contain utility routines that can be used as input to the Watcom Linker to create an application. The following are some of the advantages of using library files. 1. Only those modules that are referenced will be included in the executable file. This eliminates the need to know which object files should be included and which ones should be left out when linking an application. 2. Libraries are a good way of organizing object files. When linking an application, you need only list one library file instead of several object files. The Watcom Library Manager currently runs under the following operating systems.  DOS  OS/2  QNX  Windows ═══ 3.1. The Watcom Library Manager Command Line ═══ The following describes the Watcom Library Manager command line. WLIB [options_1] lib_file [options_2] [cmd_list] The square brackets "[]" denote items which are optional. lib_file is the file specification for the library file to be processed. If no file extension is specified, a file extension of "lib" is assumed. options_1 is a list of valid options. Options may be specified in any order. If you are using a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager, options are preceded by a "/" or "-" character. If you are using a QNX-hosted version of the Watcom Library Manager, options are preceded by a "-" character. options_2 is a list of valid options. These options are only permitted if you are running a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager and must be preceded by a "/" character. The "-" character cannot be used as an option delimiter for options following the library file name since it will be interpreted as a delete command. cmd_list is a list of commands to the Watcom Library Manager specifying what operations are to be performed. Each command in cmd_list is separated by a space. The following is a summary of valid options. Items enclosed in square brackets "[]" are optional. Items separated by an or-bar "|" and enclosed in parentheses "()" indicate that one of the items must be specified. Items enclosed in angle brackets "<>" are to be replaced with a user-supplied name or value (the "<>" are not included in what you specify). ? display the usage message b suppress creation of backup file c perform case sensitive comparison d= directory in which extracted object modules will be placed fa output AR format library fm output MLIB format library fo output OMF format library h display the usage message ia generate AXP import records ii generate X86 import records ip generate PPC import records ie generate ELF import records ic generate COFF import records io generate OMF import records i(r|n)(n|o) imports for the resident/non-resident names table are to be imported by name/ordinal. l[=] create a listing file m display C++ mangled names n always create a new library o= set output file name for library p= set library page size (supported for "OMF" library format only) q suppress identification banner s strip line number records from object files (supported for "OMF" library format only) t remove path information from module name specified in THEADR records (supported for "OMF" library format only) v do not suppress identification banner x extract all object modules from library The following sections describe the operations that can be performed on a library file. Note that before making a change to a library file, the Watcom Library Manager makes a backup copy of the original library file unless the "o" option is used to specify an output library file whose name is different than the original library file, or the "b" option is used to suppress the creation of the backup file. The backup copy has the same file name as the original library file but has a file extension of "bak". Hence, lib_file should not have a file extension of "bak". ═══ 3.2. Adding Modules to a Library File ═══ An object file can be added to a library file by specifying a +obj_file command where obj_file is the file specification for an object file. If you are using a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager, a file extension of "obj" is assumed if none is specified. If you are using a QNX-hosted version of the Watcom Library Manager, a file extension of "o" is assumed if none is specified. If the library file does not exist, a warning message will be issued and the library file will be created. Example: wlib mylib +myobj In the above example, the object file "myobj" is added to the library file "mylib.lib". When a module is added to a library, the Watcom Library Manager will issue a warning if a symbol redefinition occurs. This will occur if a symbol in the module being added is already defined in another module that already exists in the library file. Note that the module will be added to the library in any case. It is also possible to combine two library files together. The following example adds all modules in the library "newlib.lib" to the library "mylib.lib". Example: wlib mylib +newlib.lib Note that you must specify the "lib" file extension. Otherwise, the Watcom Library Manager will assume you are adding an object file. ═══ 3.3. Deleting Modules from a Library File ═══ A module can be deleted from a library file by specifying a -mod_name command where mod_name is the file name of the object file when it was added to the library with the directory and file extension removed. Example: wlib mylib -myobj In the above example, the Watcom Library Manager is instructed to delete the module "myobj" from the library file "mylib.lib". It is also possible to specify a library file instead of a module name. Example: wlib mylib -oldlib.lib In the above example, all modules in the library file "oldlib.lib" are removed from the library file "mylib.lib". Note that you must specify the "lib" file extension. Otherwise, the Watcom Library Manager will assume you are removing an object module. ═══ 3.4. Replacing Modules in a Library File ═══ A module can be replaced by specifying a -+mod_name or +-mod_name command. The module mod_name is deleted from the library. The object file "mod_name" is then added to the library. Example: wlib mylib -+myobj In the above example, the module "myobj" is replaced by the object file "myobj". It is also possible to merge two library files. Example: wlib mylib -+updlib.lib In the above example, all modules in the library file "updlib.lib" replace the corresponding modules in the library file "mylib.lib". Any module in the library "updlib.lib" not in library "mylib.lib" is added to the library "mylib.lib". Note that you must specify the "lib" file extension. Otherwise, the Watcom Library Manager will assume you are replacing an object module. ═══ 3.5. Extracting a Module from a Library File ═══ A module can be extracted from a library file by specifying a *mod_name command for a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager or a :mod_name command for a QNX-hosted version of the Watcom Library Manager. The module mod_name is not deleted but is copied to a disk file. If mod_name is preceded by a path specification, the output file will be placed in the directory identified by the path specification. If mod_name is followed by a file extension, the output file will contain the specified file extension. Example: wlib mylib *myobj DOS, OS/2 or Windows-hosted or wlib mylib :myobj QNX-hosted In the above example, the module "myobj" is copied to a disk file. The disk file will be an object file with file name "myobj". If you are running a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager, a file extension of "obj" will be used. If you are running a QNX-hosted version of the Watcom Library Manager, a file extension of "o" will be used. Example: wlib mylib *myobj.out DOS, OS/2 or Windows-hosted or wlib mylib :myobj.out QNX-hosted In the above example, the module "myobj" will be extracted from the library file "mylib.lib" and placed in the file "myobj.out" You can extract a module from a file and have that module deleted from the library file by specifying a *-mod_name command for a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager or a :-mod_name command for a QNX-hosted version of the Watcom Library Manager. The following example performs the same operations as in the previous example but, in addition, the module is deleted from the library file. Example: wlib mylib *-myobj.out DOS, OS/2 or Windows-hosted or wlib mylib :-myobj.out QNX-hosted Note that the same result is achieved if the delete operator precedes the extract operator. ═══ 3.6. Creating Import Libraries ═══ The Watcom Library Manager can also be used to create import libraries from Dynamic Link Libraries. Import libraries are used when linking OS/2, Win16 or Win32 applications. Example: wlib implib +dynamic.dll In the above example, the following actions are performed. For each external symbol in the specified Dynamic Link Library, a special object module is created that identifies the external symbol and the actual name of the Dynamic Link Library it is defined in. This object module is then added to the specified library. The resulting library is called an import library. Note that you must specify the "dll" file extension. Otherwise, the Watcom Library Manager will assume you are adding an object file. ═══ 3.7. Creating Import Library Entries ═══ An import library entry can be created and added to a library by specifying a command of the following form. ++sym.dll_name[.[altsym].export_name][.ordinal] where description sym is the name of a symbol in a Dynamic Link Library. dll_name is the name of the Dynamic Link Library that defines sym. altsym is the name of a symbol in a Dynamic Link Library. When omitted, the default symbol name is sym. export_name is the name that an application that is linking to the Dynamic Link Library uses to reference sym. When omitted, the default export name is sym. ordinal is the ordinal value that can be used to identify sym instead of using the name export_name. Example: wlib math ++__sin.trig.sin.1 In the above example, an import library entry will be created for symbol sin and added to the library "math.lib". The symbol sin is defined in the Dynamic Link Library called "trig.dll" as __sin. When an application is linked with the library "math.lib", the resulting executable file will contain an import by ordinal value 1. If the ordinal value was omitted, the resulting executable file would contain an import by name sin. ═══ 3.8. Commands from a File or Environment Variable ═══ The Watcom Library Manager can be instructed to process all commands in a disk file or environment variable by specifying the @name command where name is a file specification for the command file or the name of an environment variable. A file extension of "lbc" is assumed for files if none is specified. The commands must be one of those previously described. Example: wlib mylib @mycmd In the above example, all commands in the environment variable "mycmd" or file "mycmd.lbc" are processed by the Watcom Library Manager. ═══ 3.9. Watcom Library Manager Options ═══ The following sections describe the list of options allowed when invoking the Watcom Library Manager. ═══ 3.9.1. Suppress Creation of Backup File - "b" Option ═══ The "b" option tells the Watcom Library Manager to not create a backup library file. In the following example, the object file identified by "new" will be added to the library file "mylib.lib". Example: wlib -b mylib +new If the library file "mylib.lib" already exits, no backup library file ("mylib.bak") will be created. ═══ 3.9.2. Case Sensitive Symbol Names - "c" Option ═══ The "c" option tells the Watcom Library Manager to use a case sensitive compare when comparing a symbol to be added to the library to a symbol already in the library file. This will cause the names "myrtn" and "MYRTN" to be treated as different symbols. By default, comparisons are case insensitive. That is the symbol "myrtn" is the same as the symbol "MYRTN". ═══ 3.9.3. Specify Output Directory - "d" Option ═══ The "d" option tells the Watcom Library Manager the directory in which all extracted modules are to be placed. The default is to place all extracted modules in the current directory. In the following example, the module "mymod" is extracted from the library "mylib.lib". If you are running a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager, the module will be placed in the file "\obj\mymod.obj". If you are running a QNX-hosted version of the Watcom Library Manager, the module will be placed in the file "/o/mymod.o". Example: wlib -d=\obj mymod DOS, OS/2 or Windows-hosted or wlib -d=/o mymod QNX-hosted ═══ 3.9.4. Specify Output Format - "f" Option ═══ The "f" option tells the Watcom Library Manager the format of the output library. The default output format is determined by the type of object files that are added to the library when it is created. The possible output format options are: fa output AR format library fm output MLIB format library fo output OMF format library ═══ 3.9.5. Generating Imports - "i" Option ═══ The "i" option can be used to describe type of import library to create. ia generate AXP import records ii generate X86 import records ip generate PPC import records ie generate ELF import records ic generate COFF import records io generate OMF import records When creating import libraries from Dynamic Link Libraries, import entries for the names in the resident and non-resident names tables are created. The "i" option can be used to describe the method used to import these names. iro Specifying "iro" causes imports for names in the resident names table to be imported by ordinal. irn Specifying "irn" causes imports for names in the resident names table to be imported by name. This is the default. ino Specifying "ino" causes imports for names in the non-resident names table to be imported by ordinal. This is the default. inn Specifying "inn" causes imports for names in the non-resident names table to be imported by name. Example: wlib -iro -inn implib +dynamic.dll Note that you must specify the "dll" file extension for the Dynamic Link Library. Otherwise an object file will be assumed. ═══ 3.9.6. Creating a Listing File - "l" Option ═══ The "l" (lower case "L") option instructs the Watcom Library Manager to produce a list of the names of all symbols that can be found in the library file to a listing file. The file name of the listing file is the same as the file name of the library file. The file extension of the listing file is "lst". Example: wlib -l mylib In the above example, the Watcom Library Manager is instructed to list the contents of the library file "mylib.lib" and produce the output to a listing file called "mylib.lst". An alternate form of this option is -l=list_file. With this form, you can specify the name of the listing file. When specifying a listing file name, a file extension of "lst" is assumed if none is specified. Example: wlib -l=mylib.out mylib In the above example, the Watcom Library Manager is instructed to list the contents of the library file "mylib.lib" and produce the output to a listing file called "mylib.out". You can get a listing of the contents of a library file to the terminal by specifying only the library name on the command line as demonstrated by the following example. Example: wlib mylib ═══ 3.9.7. Display C++ Mangled Names - "m" Option ═══ The "m" option instructs the Watcom Library Manager to display C++ mangled names rather than displaying their demangled form. The default is to interpret mangled C++ names and display them in a somewhat more intelligible form. ═══ 3.9.8. Always Create a New Library - "n" Option ═══ The "n" option tells the Watcom Library Manager to always create a new library file. If the library file already exists, a backup copy is made (unless the "b" option was specified). The original contents of the library are discarded and a new library is created. If the "n" option was not specified, the existing library would be updated. Example: wlib -n mylib +myobj In the above example, a library file called "mylib.lib" is created. It will contain a single object module, namely "myobj", regardless of the contents of "mylib.lib" prior to issuing the above command. If "mylib.lib" already exists, it will be renamed to "mylib.bak". ═══ 3.9.9. Specifying an Output File Name - "o" Option ═══ The "o" option can be used to specify the output library file name if you want the original library to remain unchanged and a new library created. Example: wlib -o=newlib lib1 +lib2.lib In the above example, the modules from "lib1.lib" and "lib2.lib" are added to the library "newlib.lib". Note that since the original library remains unchanged, no backup copy is created. Also, if the "l" option is used to specify a listing file, the listing file will assume the file name of the output library. ═══ 3.9.10. Specifying a Library Record Size - "p" Option ═══ The "p" option specifies the record size in bytes for each record in the library file. The record size must be a power of 2 and in the range 16 to 32768. If the record size is less than 16, it will be rounded up to 16. If the record size is greater than 16 and not a power of 2, it will be rounded up to the nearest power of 2. The default record size is 256 bytes. Each entry in the dictionary of a library file contains an offset from the start of the file which points to a module. The offset is 16 bits and is a multiple of the record size. Since the default record size is 256, the maximum size of a library file for a record size of 256 is 256*64K. If the size of the library file increases beyond this size, you must increase the record size. Example: wlib -p=512 lib1 +lib2.lib In the above example, the Watcom Library Manager is instructed to create/update the library file "lib1.lib" by adding the modules from the library file "lib2.lib". The record size of the resulting library file is 512 bytes. ═══ 3.9.11. Operate Quietly - "q" Option ═══ The "q" option suppressing the banner and copyright notice that is normally displayed when the Watcom Library Manager is invoked. Example: wlib -q -l mylib ═══ 3.9.12. Strip Line Number Records - "s" Option ═══ The "s" option tells the Watcom Library Manager to remove line number records from object files that are being added to a library. Line number records are generated in the object file if the "d1" option is specified when compiling the source code. Example: wlib -s mylib +myobj ═══ 3.9.13. Trim Module Name - "t" Option ═══ The "t" option tells the Watcom Library Manager to remove path information from the module name specified in THEADR records in object files that are being added to a library. The module name is created from the file name by the compiler and placed in the THEADR record of the object file. The module name will contain path information if the file name given to the compiler contains path information. Example: wlib -t mylib +myobj ═══ 3.9.14. Operate Verbosely - "v" Option ═══ The "v" option enables the display of the banner and copyright notice when the Watcom Library Manager is invoked. Example: wlib -v -l mylib ═══ 3.9.15. Explode Library File - "x" Option ═══ The "x" option tells the Watcom Library Manager to extract all modules from the library. Note that the modules are not deleted from the library. Object modules will be placed in the current directory unless the "d" option is used to specify an alternate directory. In the following example all modules will be extracted from the library "mylib.lib" and placed in the current directory. Example: wlib -x mylib In the following example, all modules will be extracted from the library "mylib.lib". If you are running a DOS, OS/2 or Windows-hosted version of the Watcom Library Manager, the module will be placed in the "\obj" directory. If you are running a QNX-hosted version of the Watcom Library Manager, the module will be placed in the file "/o" directory. Example: wlib -x -d=\obj mylib DOS, OS/2 or Windows-hosted or wlib -x -d=/o mylib QNX-hosted ═══ 3.10. Librarian Error Messages ═══ The following messages may be issued by the Watcom Library Manager. Error! Could not open object file '%s'. Object file '%s' could not be found. This message is usually issued when an attempt is made to add a non-existent object file to the library. Error! Could not open library file '%s'. The specified library file could not be found. This is usually issued for input library files. For example, if you are combining two library files, the library file you are adding is an input library file and the library file you are adding to or creating is an output library file. Error! Invalid object module in file '%s' not added. The specified file contains an invalid object module. Error! Dictionary too large. Recommend split library into two libraries. The size of the dictionary in a library file cannot exceed 64K. You must split the library file into two separate library files. Error! Redefinition of module '%s' in file '%s'. This message is usually issued when an attempt is made to add a module to a library that already contains a module by that name. Warning! Redefinition of symbol '%s' in file '%s' ignored. This message is issued if a symbol defined by a module already in the library is also defined by a module being added to the library. Error! Library too large. Recommend split library into two libraries or try a larger page_bound than %xH. The record size of the library file does not allow the library file to increase beyond its current size. The record size of the library file must be increased using the "p" option. Error! Expected '%s' in '%s' but found '%s'. An error occurred while scanning command input. Warning! Could not find module '%s' for deletion. This message is issued if an attempt is made to delete a module that does not exist in the library. Error! Could not find module '%s' for extraction. This message is issued if an attempt is made to extract a module that does not exist in the library. Error! Could not rename old library for backup. The Watcom Library Manager creates a backup copy before making any changes (unless the "b" option is specified). This message is issued if an error occurred while trying to rename the original library file to the backup file name. Warning! Could not open library '%s' : will be created. The specified library does not exist. It is usually issued when you are adding to a non-existent library. The Watcom Library Manager will create the library. Warning! Output library name specification ignored. This message is issued if the library file specified by the "o" option could not be opened. Warning! Could not open library '%s' and no operations specified: will not be created. This message is issued if the library file specified on the command line does not exist and no operations were specified. For example, asking for a listing file of a non-existent library will cause this message to be issued. Warning! Could not open listing file '%s'. The listing file could not be opened. For example, this message will be issued when a "disk full" condition is present. Error! Could not open output library. The output library could not be opened. Error! Unable to write to output library. An error occurred while writing to the output library. Error! Unable to write to extraction file '%s'. This message is issued when extracting an object module from a library file and an error occurs while writing to the output file. Error! Out of Memory. There was not enough memory to process the library file. Error! Could not open file '%s'. This message is issued if the output file for a module that is being extracted from a library could not be opened. Error! Library '%s' is invalid. Contents ignored. The library file does not contain the correct header information. Error! Library '%s' has an invalid page size. Contents ignored. The library file has an invalid record size. The record size is contained in the library header and must be a power of 2. Error! Invalid object record found in file '%s'. The specified file contains an invalid object record. Error! No library specified on command line. This message is issued if a library file name is not specified on the command line. Error! Expecting library name. This message is issued if the location of the library file name on the command line is incorrect. Warning! Invalid file name '%s'. This message is issued if an invalid file name is specified. For example, a file name longer that 127 characters is not allowed. Error! Could not open command file '%s'. The specified command file could not be opened. Error! Could not read from file '%s'. Contents ignored as command input. An error occurred while reading a command file. ═══ 4. The Object File Disassembler ═══ This chapter describes the Watcom Disassembler. It takes as input an object file (a file with extension ".obj") and produces, as output, the Intel assembly language equivalent. The Watcom compilers do not produce an assembly language listing directly from a source program. Instead, the Watcom Disassembler can be used to generate an assembly language listing from the object file generated by the compiler. The Watcom Disassembler command line syntax is the following. WDIS [options] [d:][path]filename[.ext] [options] The square brackets [ ] denote items which are optional. WDIS is the name of the Watcom Disassembler. d: is an optional drive specification such as "A:", "B:", etc. If not specified, the default drive is assumed. path is an optional path specification such as "\PROGRAMS\OBJ\". If not specified, the current directory is assumed. filename is the file name of the object file to disassemble. ext is the file extension of the object file to disassemble. If omitted, a file extension of ".obj" is assumed. If the period "." is specified but not the extension, the file is assumed to have no file extension. options is a list of valid options, each preceded by a slash ("/") or a dash ("-"). Options may be specified in any order. The options supported by the Watcom Disassembler are: a write assembly instructions only to the listing file e include list of external names fp do not use instruction name pseudonyms fr do not use register name pseudonyms [Alpha only] fi use alternate indexing format [80(x)86 only] fu instructions/registers in upper case i= redefine the initial character of internal labels (default: L) l[=] create a listing file m leave C++ names mangled o print list of operands beside instructions p include list of public names s[=] using object file source line information, imbed original source lines into the output file The following sections describe the list of options. ═══ 4.1. Changing the Internal Label Character - "i=" ═══ The "i" option permits you to specify the first character to be used for internal labels. Internal labels take the form "Ln" where "n" is one or more digits. The default character "L" can be changed using the "i" option. The replacement character must be a letter (a-z, A-Z). A lowercase letter is converted to uppercase. Example: C>wdis calendar /i=x ═══ 4.2. The Assembly Format Option - "a" ═══ The "a" option controls the format of the output produced to the listing file. When specified, the Watcom Disassembler will produce a listing file that can be used as input to an assembler. Example: C>wdis calendar /a /l=calendar.asm In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file CALENDAR.OBJ and produce the output to the file CALENDAR.ASM so that it can be assembled by an assembler. ═══ 4.3. The External Symbols Option - "e" ═══ The "e" option controls the amount of information produced in the listing file. When specified, a list of all externally defined symbols is produced in the listing file. Example: C>wdis calendar /e In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file CALENDAR.OBJ and produce the output, with a list of all external symbols, on the screen. A sample list of external symbols is shown below. List of external symbols Symbol ---------------- ___iob 0000032f 00000210 000001f4 00000158 00000139 __CHK 00000381 00000343 000002eb 00000237 000000cb 00000006 Box_ 000000f2 Calendar_ 000000a7 00000079 00000049 ClearScreen_ 00000016 fflush_ 00000334 00000215 000001f9 0000015d 0000013e int386_ 000003af 00000372 Line_ 000002db 000002b5 00000293 00000274 0000025a localtime_ 00000028 memset_ 00000308 PosCursor_ 0000031e 000001e1 00000148 00000123 000000b6 printf_ 00000327 00000208 000001ec 00000150 00000131 strlen_ 00000108 time_ 0000001d ------------------------------------------------------------ Each externally defined symbol is followed by a list of location counter values indicating where the symbol is referenced. The "e" option is ignored when the "a" option is specified. ═══ 4.4. The No Instruction Name Pseudonyms Option - "fp" ═══ By default, AXP instruction name pseudonyms are emitted in place of actual instruction names. The Watcom AXP Assembler accepts instruction name pseudonyms. The "fp" option instructs the Watcom Disassembler to emit the actual instruction names instead. ═══ 4.5. The No Register Name Pseudonyms Option - "fr" ═══ By default, AXP register names are emitted in pseudonym form. The Watcom AXP Assembler accepts register pseudonyms. The "fr" option instructs the Watcom Disassembler to display register names in their non-pseudonym form. ═══ 4.6. The Alternate Addressing Form Option - "fi" ═══ The "fi" option causes an alternate syntactical form of the based or indexed addressing mode of the 80x86 to be used in an instruction. For example, the following form is used by default for Intel instructions. mov ax,-2[bp] If the "fi" option is specified, the following form is used. mov ax,[bp-2] ═══ 4.7. The Uppercase Instructions/Registers Option - "fu" ═══ The "fu" option instructs the Watcom Disassembler to display instruction and register names in uppercase characters. The default is to display them in lowercase characters. ═══ 4.8. The Listing Option - "l[=]" ═══ By default, the Watcom Disassembler produces its output to the terminal. The "l" (lowercase L) option instructs the Watcom Disassembler to produce the output to a listing file. The default file name of the listing file is the same as the file name of the object file. The default file extension of the listing file is .LST. Example: C>wdis calendar /l In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file CALENDAR.OBJ and produce the output to a listing file called CALENDAR.LST. An alternate form of this option is "l=". With this form, you can specify the name of the listing file. When specifying a listing file, a file extension of .LST is assumed if none is specified. Example: C>wdis calendar /l=calendar.lis In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file CALENDAR.OBJ and produce the output to a listing file called CALENDAR.LIS. ═══ 4.9. The Public Symbols Option - "p" ═══ The "p" option controls the amount of information produced in the listing file. When specified, a list of all public symbols is produced in the listing file. Example: C>wdis calendar /p In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file CALENDAR.OBJ and produce the output, with a list of all exported symbols, to the screen. A sample list of public symbols is shown below. The following is a list of public symbols in 80x86 code. List of public symbols SYMBOL SECTION OFFSET -------------------------------------------------------- main_ _TEXT 000002C0 void near Box( int, int, int, int ) _TEXT 00000093 void near Calendar( int, int, int, int, int, char near * ) _TEXT 0000014A void near ClearScreen() _TEXT 00000000 void near Line( int, int, int, char, char, char ) _TEXT 00000036 void near PosCursor( int, int ) _TEXT 0000001A The following is a list of public symbols in Alpha AXP code. List of public symbols SYMBOL SECTION OFFSET -------------------------------------------------------- main .text 000004F0 void near Box( int, int, int, int ) .text 00000148 void near Calendar( int, int, int, int, int, char near * ) .text 00000260 void near ClearScreen() .text 00000000 void near Line( int, int, int, char, char, char ) .text 00000060 void near PosCursor( int, int ) .text 00000028 The "p" option is ignored when the "a" option is specified. ═══ 4.10. Retain C++ Mangled Names - "m" ═══ The "m" option instructs the Watcom Disassembler to retain C++ mangled names rather than displaying their demangled form. The default is to interpret mangled C++ names and display them in a somewhat more intelligible form. ═══ 4.11. The Source Option - "s[=]" ═══ The "s" option causes the source lines corresponding to the assembly language instructions to be produced in the listing file. The object file must contain line numbering information. That is, the "d1" or "d2" option must have been specified when the source file was compiled. If no line numbering information is present in the object file, the "s" option is ignored. The following defines the order in which the source file name is determined when the "s" option is specified. 1. If present, the source file name specified on the command line. 2. The name from the module header record. 3. The object file name. In the following example, we have compiled the source file MYSRC.C with "d1" debugging information. We then disassemble it as follows: Example: C>wdis mysrc /s /l In the above example, the Watcom Disassembler is instructed to disassemble the contents of the file MYSRC.OBJ and produce the output to the listing file MYSRC.LST. The source lines are extracted from the file MYSRC.C. An alternate form of this option is "s=". With this form, you can specify the name of the source file. Example: C>wdis mysrc /s=myprog.c /l The above example produces the same result as in the previous example except the source lines are extracted from the file MYPROG.C. ═══ 4.12. An Example ═══ Consider the following program contained in the file HELLO.C. ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── #include void main() { printf( "Hello world\n" ); } ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Compile it with the "d1" option. An object file called HELLO.OBJ will be produced. The "d1" option causes line numbering information to be generated in the object file. We can use the Watcom Disassembler to disassemble the contents of the object file by issuing the following command. C>wdis hello /l /e /p /s /fu The output will be written to a listing file called HELLO.LST (the "l" option was specified"). It will contain a list of external symbols (the "e" option was specified), a list of public symbols (the "p" option was specified) and the source lines corresponding to the assembly language instructions (the "s" option was specified). The source input file is called HELLO.C. The register names will be displayed in upper case (the "fu" option was specified). The output, shown below, is the result of using the Watcom C++ compiler. The following is a disassembly of 80x86 code. Module: HELLO.C GROUP: 'DGROUP' CONST,CONST2,_DATA,_BSS Segment: _TEXT DWORD USE32 0000001A bytes #include void main() 0000 main_: 0000 68 08 00 00 00 PUSH 0x00000008 0005 E8 00 00 00 00 CALL __CHK { printf( "Hello world\n" ); 000A 68 00 00 00 00 PUSH offset L$1 000F E8 00 00 00 00 CALL printf_ 0014 83 C4 04 ADD ESP,0x00000004 } 0017 31 C0 XOR EAX,EAX 0019 C3 RET Routine Size: 26 bytes, Routine Base: _TEXT + 0000 No disassembly errors List of external references SYMBOL ------- __CHK 0006 printf_ 0010 Segment: CONST DWORD USE32 0000000D bytes 0000 L$1: 0000 48 65 6C 6C 6F 20 77 6F 72 6C 64 0A 00 Hello world.. BSS Size: 0 bytes List of public symbols SYMBOL SECTION OFFSET -------------------------------------------------------- main_ _TEXT 00000000 The following is a disassembly of Alpha AXP code. .new_section .text, "crx4" #include void main() 0000 main: 0000 23DEFFF0 LDA SP,-0x10(SP) 0004 B75E0000 STQ RA,(SP) { printf( "Hello world\n" ); 0008 261F0000 LDAH A0,h^L$0(R31) 000C 22100000 LDA A0,l^L$0(A0) 0010 43F00010 SEXTL A0,A0 0014 D3400000 BSR RA,j^printf } 0018 201F0000 MOV 0x00000000,V0 001C A75E0000 LDQ RA,(SP) 0020 23DE0010 LDA SP,0x10(SP) 0024 6BFA8001 RET (RA) Routine Size: 40 bytes, Routine Base: .text + 0000 No disassembly errors List of external references SYMBOL ------- printf 0014 .new_section .const, "drw4" 0000 L$0: 0000 48 65 6C 6C 6F 20 77 6F 72 6C 64 0A 00 00 00 00 Hello world..... .new_section .const2, "drw4" .new_section .data, "drw4" .new_section .bss, "urw4" 0000 .bss: BSS Size: 0 bytes .new_section .pdata, "dr2" 0000 // Procedure descriptor for main main // BeginAddress : 0 main+0x28 // EndAddress : 40 00000000 // ExceptionHandler : 0 00000000 // HandlerData : 0 main+0x8 // PrologEnd : 8 .new_section .drectve, "iRr0" 0000 2D 64 65 66 61 75 6C 74 6C 69 62 3A 63 6C 69 62 -defaultlib:clib 0010 20 2D 64 65 66 61 75 6C 74 6C 69 62 3A 70 6C 69 -defaultlib:pli 0020 62 20 2D 64 65 66 61 75 6C 74 6C 69 62 3A 6D 61 b -defaultlib:ma 0030 74 68 20 00 th . List of public symbols SYMBOL SECTION OFFSET -------------------------------------------------------- main .text 00000000 Let us create a form of the listing file that can be used as input to an assembler. C>wdis hello /l=hello.asm /r /a The output will be produced in the file HELLO.ASM. The output, shown below, is the result of using the Watcom C++ compiler. The following is a disassembly of 80x86 code. .387 .386p PUBLIC main_ EXTRN __CHK:BYTE EXTRN printf_:BYTE EXTRN ___wcpp_3_data_init_fs_root_:BYTE EXTRN _cstart_:BYTE DGROUP GROUP CONST,CONST2,_DATA,_BSS _TEXT SEGMENT DWORD PUBLIC USE32 'CODE' ASSUME CS:_TEXT, DS:DGROUP, SS:DGROUP main_: PUSH 0x00000008 CALL near ptr __CHK PUSH offset L$1 CALL near ptr printf_ ADD ESP,0x00000004 XOR EAX,EAX RET _TEXT ENDS CONST SEGMENT DWORD PUBLIC USE32 'DATA' L$1: DB 0x48, 0x65, 0x6c, 0x6c, 0x6f, 0x20, 0x77, 0x6f DB 0x72, 0x6c, 0x64, 0x0a, 0x00 CONST ENDS CONST2 SEGMENT DWORD PUBLIC USE32 'DATA' CONST2 ENDS _DATA SEGMENT DWORD PUBLIC USE32 'DATA' _DATA ENDS _BSS SEGMENT DWORD PUBLIC USE32 'BSS' _BSS ENDS END The following is a disassembly of Alpha AXP code. .globl main .extrn printf .extrn _cstart_ .new_section .text, "crx4" main: LDA $SP,-0x10($SP) STQ $RA,($SP) LDAH $A0,h^`L$0`($ZERO) LDA $A0,l^`L$0`($A0) SEXTL $A0,$A0 BSR $RA,j^printf MOV 0x00000000,$V0 LDQ $RA,($SP) LDA $SP,0x10($SP) RET $ZERO,($RA),0x00000001 .new_section .const, "drw4" `L$0`: .asciiz "Hello world\n" .byte 0x00, 0x00, 0x00 .new_section .pdata, "dr2" // 0000 Procedure descriptor for main .long main // BeginAddress : 0 .long main+0x28 // EndAddress : 40 .long 00000000 // ExceptionHandler : 0 .long 00000000 // HandlerData : 0 .long main+0x8 // PrologEnd : 8 .new_section .drectve, "iRr0" .asciiz "-defaultlib:clib -defaultlib:plib -defaultlib:math " ═══ 5. Optimization of Far Calls ═══ Optimization of far calls can result in smaller executable files and improved performance. It is most useful when the automatic grouping of logical segments into physical segments takes place. Note that, by default, automatic grouping is performed by the Watcom Linker. The Watcom C, C++ and FORTRAN 77 compilers automatically enable the far call optimization. The Watcom Linker will optimize far calls to procedures that reside in the same physical segment as the caller. For example, a large code model program will probably contain many far calls to procedures in the same physical segment. Since the segment address of the caller is the same as the segment address of the called procedure, only a near call is necessary. A near call does not require a relocation entry in the relocation table of the executable file whereas a far call does. Thus, the far call optimization will result in smaller executable files that will load faster. Furthermore, a near call will generally execute faster than a far call, particularly on 286 and 386-based machines where, for applications running in protected mode, segment switching is fairly expensive. The following describes the far call optimization. The call far label instruction is converted to one of the following sequences of code. push cs seg ss call near label push cs nop call near label Notes: 1. The nop or seg ss instruction is present since a call far label instruction is five bytes. The push cs instruction is one byte and the call near label instruction is three bytes. The seg ss instruction is used because it is faster than the nop instruction. 2. The called procedure will still use a retf instruction but since the code segment and the near address are pushed on the stack, the far return will execute correctly. 3. The position of the padding instruction is chosen so that the return address is word aligned. A word aligned return address improves performance. 4. When two consecutive call far label instructions are optimized and the first call far label instruction is word aligned, the following sequence replaces both call far label instructions. push cs call near label1 seg ss push cs seg cs call near label2 5. If your program contains only near calls, this optimization will have no effect. A far jump optimization is also performed by the Watcom Linker. This has the same benefits as the far call optimization. A jmp far label instruction to a location in the same segment will be replaced by the following sequence of code. jmp near label mov ax,ax Note that for 32-bit segments, this instruction becomes mov eax,eax. ═══ 5.1. Far Call Optimizations for Non-Watcom Object Modules ═══ The far call optimization is automatically enabled when object modules created by the Watcom C, C++, or FORTRAN 77 compilers are linked. These compilers mark those segments in which this optimization can be performed. The following utility can be used to enable this optimization for object modules that have been created by other compilers or assemblers. ═══ 5.1.1. The Watcom Far Call Optimization Enabling Utility ═══ Only DOS, OS/2 and Windows-hosted versions of the Watcom Far Call Optimization Enabling Utility are available. A QNX-hosted version is not necessary since QNX-hosted development tools that generate object files, generate the necessary information that enables the far call optimization. The format of the Watcom Far Call Optimization Enabling Utility is as follows. Items enclosed in square brackets are optional; items enclosed in braces may be repeated zero or more times. FCENABLE { [option] [file] } where description option is an option and must be preceded by a dash ('-') or slash ('/'). file is a file specification for an object file or library file. If no file extension is specified, a file extension of "obj" is assumed. Wild card specifiers may be used. The following describes the command line options. b Do not create a backup file. By default, a backup file will be created. The backup file name will have the same file name as the input file and a file extension of "bob" for object files and "bak" for library files. c Specify a list of class names, each separated by a comma. This enables the far call optimization for all segments belonging to the specified classes. s Specify a list of segment names, each separated by a comma. This enables the far call optimization for all specified segments. x Specify a list of ranges, each separated by a comma, for which no far call optimizations are to be made. A range has the following format. seg_name start-end or seg_name start:length seg_name is the name of a segment. start is an offset into the specified segment defining the start of the range. end is an offset into the specified segment defining the end of the range. length is the number of bytes from start to be included in the range. All values are assumed to be hexadecimal. Notes: 1. If more than one class list or segment list is specified, only the last one is used. A class or segment list applies to all object and library files regardless of their position relative to the class or segment list. 2. A range list applies only to the first object file following the range specification. If the object file contains more than one module, the range list will only apply to the first module in the object file. The following examples illustrate the use of the Watcom Far Call Optimization Enabling Utility. Example: fcenable /c code *.obj In the above example, the far call optimization will be enabled for all segments belonging to the "code" class. Example: fcenable /s _text *.obj In the above example, the far call optimization will be enabled for all segments with name "_text". Example: fcenable /x special 0:400 asmfile.obj In the above example, the far call optimization will be disabled for the first 1k bytes of the segment named "special" in the object file "asmfile". Example: fcenable /x special 0-ffffffff asmfile.obj In the above example, the far call optimization will be disabled for the entire segment named "special" in the object file "asmfile". ═══ 6. The Watcom Patch Utility ═══ The Watcom Patch Utility is a utility program which may be used to apply patches or bug fixes to Watcom's compilers and its associated tools. As problems are reported and fixed, patches are created and made available on Watcom's BBS, Watcom's FTP site, or CompuServe for users to download and apply to their copy of the tools. ═══ 6.1. Applying a Patch ═══ The format of the BPATCH command line is: BPATCH [options] patch_file The square brackets [ ] denote items which are optional. where description options is a list of valid Watcom Patch Utility options, each preceded by a dash ("-"). Options may be specified in any order. The possible options are: -p Do not prompt for confirmation -b Do not create a .BAK file -q Print current patch level of file patch_file is the file specification for a patch file provided by Watcom. Suppose a patch file called "wlink.a" is supplied by Watcom to fix a bug in the file "WLINK.EXE". The patch may be applied by typing the command: bpatch wlink.a The Watcom Patch Utility locates the file C:\WATCOM\BINW\WLINK.EXE using the PATH environment variable. The actual name of the executable file is extracted from the file WLINK.A. It then verifies that the file to be patched is the correct one by comparing the size of the file to be patched to the expected size. If the file sizes match, the program responds with: Ok to modify 'C:\WATCOM\BINW\WLINK.EXE'? [y|n] If you respond with "yes", BPATCH will modify the indicated file. If you respond with "no", BPATCH aborts. Once the patch has been applied the resulting file is verified. First the file size is checked to make sure it matches the expected file size. If the file size matches, a check-sum is computed and compared to the expected check-sum. Notes: 1. If an error message is issued during the patch process, the file that you specified to be patched will remain unchanged. 2. If a sequence of patch files exist, such as "wlink.a", "wlink.b" and "wlink.c", the patches must be applied in order. That is, "wlink.a" must be applied first followed by "wlink.b" and finally "wlink.c". ═══ 6.2. Diagnostic Messages ═══ If the patch cannot be successfully applied, one of the following error messages will be displayed. Usage: BPATCH {-p} {-q} {-b} -p = Do not prompt for confirmation -b = Do not create a .BAK file -q = Print current patch level of file The command line was entered with no arguments. File '%s' has not been patched This message is issued when the "-q" option is used and the file has not been patched. File '%s' has been patched to level '%s' This message is issued when the "-q" option is used and the file has been patched to the indicated level. File '%s' has already been patched to level '%s' - skipping This message is issued when the file has already been patched to the same level or higher. Command line may only contain one file name More than one file name is specified on the command line. Make sure that "/" is not used as an option delimiter. Command line must specify a file name No file name has been specified on the command line. '%s' is not a Watcom patch file The patch file is not of the required format. The required header information is not present. '%s' is not a valid Watcom patch file The patch file is not of the required format. The required header information is present but the remaining contents of the file have been corrupted. '%s' is the wrong size (%lu1). Should be (%lu2) The size of the file to be patched (%lu1) is not the same as the expected size (%lu2). Cannot find '%s' Cannot find the executable to be patched. Cannot open '%s' An error occurred while trying to open the patch file, the file to be patched or the resulting file. Cannot read '%s' An input error occurred while reading the old version of the file being patched. Cannot rename '%s' to '%s' The file to be patched could not be renamed to the backup file name or the resulting file could not be renamed to the name of the file that was patched. Cannot write to '%s' An output error occurred while writing to the new version of the file to be patched. I/O error processing file '%s' An error occurred while seeking in the specified file. No memory for %s An attempt to allocate memory dynamically failed. Patch program aborted! This message is issued if you answered no to the "OK to modify" prompt. Resulting file has wrong checksum (%lu) - Should be (%lu2) The check-sum of the resulting file (%lu) does not match the expected check-sum (%lu2). This message is issued if you have patched the wrong version. Resulting file has wrong size (%lu1) - Should be (%lu2) The size of the resulting file (%lu1) does not match the expected size (%lu2). This message is issued if you have patched the wrong version. ═══ 7. The Watcom Strip Utility ═══ The Watcom Strip Utility may be used to manipulate information that is appended to the end of an executable file. The information can be either one of two things: 1. Symbolic debugging information 2. Resource information This information can be added or removed from the executable file. Symbolic debugging information is placed at the end of an executable file by the Watcom Linker or the Watcom Strip Utility. Resource information is placed at the end of an executable by a resource compiler or the Watcom Strip Utility. Once a program has been debugged, the Watcom Strip Utility allows you to remove the debugging information from the executable file so that you do not have to remove the debugging directives from the linker directive file and link your program again. Removal of the debugging information reduces the size of the executable image. All executable files generated by the Watcom Linker can be specified as input to the Watcom Strip Utility. Note that for executable files created for Novell's NetWare 386 operating system, debugging information created using the "NOVELL" option in the "DEBUG" directive cannot be removed from the executable file. You must remove the "DEBUG" directive from the directive file and re-link your application. The Watcom Strip Utility currently runs under the following operating systems.  DOS  OS/2  QNX  Windows NT  Windows 95 ═══ 7.1. The Watcom Strip Utility Command Line ═══ The Watcom Strip Utility command line syntax is: WSTRIP [options] input_file [output_file] [info_file] where: [] The square brackets denote items which are optional. options /n (noerrors) Do not issue any diagnostic message. /q (quiet) Do not print any informational messages. /r (resources) Process resource information rather than debugging information. /a (add) Add information rather than remove information. input_file is a file specification for the name of an executable file. If no file extension is specified, the Watcom Strip Utility will assume one of the following extensions: "exe", "dll", "exp", "rex", "nlm", "dsk", "lan", "nam", "qnx" or no file extension. Note that the order specified in the list of file extensions is the order in which the Watcom Strip Utility will select file extensions. output_file is an optional file specification for the output file. If no file extension is specified, the file extension specified in the input file name will be used for the output file name. If "." is specified, the input file name will be used. info_file is an optional file specification for the file in which the debugging or resource information is to be stored (when removing information) or read (when adding information). If no file extension is specified, a file extension of "sym" is assumed for debugging information and "res" for resource information. To specify the name of the information file but not the name of an output file, a "." may be specified in place of output_file. Description: 1. If the "r" (resource) option is not specified then the default action is to add/remove symbolic debugging information. 2. If the "a" (add) option is not specified then the default action is to remove information. 3. If output_file is not specified, the debugging or resource information is added to or removed from input_file. 4. If output_file is specified, input_file is copied to output_file and the debugging or resource information is added to or removed from output_file. input_file remains unchanged. 5. If info_file is specified then the debugging or resource information that is added to or removed from the executable file is read from or written to this file. The debugging or resource information may be appended to the executable by specifying the "a" (add) option. Also, the debugging information may be appended to the executable by concatenating the debugging information file to the end of the executable file (the files must be treated as binary files). 6. During processing, the Watcom Strip Utility will create a temporary file, ensuring that a file by the chosen name does not already exist. ═══ 7.2. Strip Utility Messages ═══ The following messages may be issued by the Watcom Strip Utility. Usage: WSTRIP [options] input_file [output_file] [info_file] options: (-option is also accepted) /n don't print warning messages /q don't print informational messages /r process resource information rather than debugging information /a add information rather than delete information input_file: executable file output_file: optional output executable or '.' info_file: optional output debugging or resource information file or input debugging or resource informational file The command line was entered with no arguments. Too low on memory There is not enough free memory to allocate file buffers. Unable to find '%s' The specified file could not be located. Cannot create temporary file All the temporary file names are in use. Unable to open '%s' to read The input executable file cannot be opened for reading. '%s' is not a valid executable file The input file has invalid executable file header information. '%s' does not contain debugging information There is nothing to strip from the specified executable file. Seek error on '%s' An error occurred during a seek operation on the specified file. Unable to create output file '%s' The output file could not be created. Check that the output disk is not write-protected or that the specified output file is not marked "read-only". Unable to create symbol file '%s' The symbol file could not be created. Error reading '%s' An error occurred while reading the input executable file. Error writing to '%s' An error occurred while writing the output executable file or the symbol file. Check the amount of free space on the output disk. If the input and output files reside on the same disk, there might not be enough room for a second copy of the executable file during processing. Cannot erase file '%s' The input executable file is probably marked "read-only" and therefore could not be erased (the input file is erased whenever the output file has the same name). Cannot rename file '%s' The output executable file could not be renamed. Ordinarily, this should never occur. ═══ 8. The Watcom Make Utility ═══ The Watcom Make utility is useful in the development of programs and text processing but is general enough to be used in many different applications. Make uses the fact that each file has a time-stamp associated with it that indicates the last time the file was updated. Make uses this time-stamp to decide which files are out of date with respect to each other. For instance, if we have an input data file and an output report file we would like the output report file to accurately reflect the contents of the input data file. In terms of time-stamps, we would like the output report to have a more recent time-stamp than the input data file (we will say that the output report file should be "younger" than the input data file). If the input file had been modified then we would know from the younger time-stamp (in comparison to the report file) that the report file was out of date and should be updated. Make may be used in this and many other situations to ensure that files are kept up to date. Some readers will be quite familiar with the concepts of the Make file maintenance tool. Watcom Make is patterned after the Make utility found on UNIX systems. The next major section is simply intended to summarize, for reference purposes only, the syntax and options of Make's command line and special macros. Subsequent sections go into the philosophy and capabilities of Watcom Make. If you are not familiar with the capabilities of the Make utility, we recommend that you skip to the next major section entitled "Dependency Declarations" and read on. ═══ 8.1. Watcom Make Reference ═══ The following sub-sections serve as a reference guide to the Watcom Make utility. ═══ 8.1.1. Watcom Make Command Line Format ═══ The formal Watcom Make command line syntax is shown below. WMAKE [options] [macro_defs] [targets] As indicated by the square brackets [ ], all items are optional. options is a list of valid Watcom Make options, each preceded by a slash ("/") or a dash ("-"). Options may be specified in any order. macro_defs is a list of valid Watcom Make macro definitions. Macro definitions are of the form: A=B and are readily identified by the presence of the "=" (the "#" character may be used instead of the "=" character if necessary). Surround the definition with quotes (") if it contains blanks (e.g., "debug_opt=debug all"). The macro definitions specified on the command line supersede any macro definitions defined in makefiles. targets is one or more targets described in the makefile. ═══ 8.1.2. Watcom Make Options Summary ═══ In this section, we present a terse summary of the Watcom Make options. This summary is displayed on the screen by simply entering "WMAKE ?" on the command line. Example: C>wmake ? /a make all targets by ignoring time-stamps /b block/ignore all implicit rules /c do not verify the existence of files made /d debug mode - echo all work as it progresses /e always erase target after error/interrupt (disables prompting) /f the next parameter is a name of dependency description file /h do not print out Make identification lines (no header) /i ignore return status of all commands executed /k on error/interrupt: continue on next target /l the next parameter is the name of a output log file /m do not search for MAKEINIT file /ms Microsoft NMAKE mode /n no execute mode - print commands without executing /o use circular implicit rule path /p print the dependency tree as understood from the file /q query mode - check targets without updating them /r do not use default definitions /s silent mode - do not print commands before execution /t touch files instead of executing commands /u UNIX compatibility mode /z do not erase target after error/interrupt (disables prompting) ═══ 8.1.3. Command Line Options ═══ Command line options, available with Watcom Make, allow you to control the processing of the makefile. ═══ 8.1.3.1. a ═══ make all targets by ignoring time-stamps The "a" option is a safe way to update every target. For program maintenance, it is the preferred method over deleting object files or touching source files. ═══ 8.1.3.2. b ═══ block/ignore all implicit rules The "b" option will indicate to Make that you do not want any implicit rule checking done. The "b" option is useful in makefiles containing double colon "::" explicit rules because an implicit rule search is conducted after a double colon "::" target is updated. Including the directive .BLOCK in a makefile also will disable implicit rule checking. ═══ 8.1.3.3. c ═══ do not verify the existence of files made Make will check to ensure that a target exists after the associated command list is executed. The target existence checking may be disabled with the "c" option. The "c" option is useful in processing makefiles that were developed with other Make utilities. The .NOCHECK directive may be used to disable target existence in a makefile. ═══ 8.1.3.4. d ═══ debug mode - echo all work as it progresses The "d" option will print out information about the time-stamp of files and indicate how the makefile processing is proceeding. ═══ 8.1.3.5. e ═══ always erase target after error/interrupt (disables prompting) The "e" option will indicate to Make that, if an error or interrupt occurs during makefile processing, the current target being made may be deleted without prompting. The .ERASE directive may be used as an equivalent option in a makefile. ═══ 8.1.3.6. f ═══ the next parameter is a name of dependency description file The "f" option specifies that the next parameter on the command line is the name of a makefile which must be processed. If the "f" option is specified then the search for the default makefile named "MAKEFILE" is not done. Any number of makefiles may be processed with the "f" option. Example: wmake /f myfile wmake /f myfile1 /f myfile2 ═══ 8.1.3.7. h ═══ do not print out Make identification lines (no header) The "h" option is useful for less verbose output. Combined with the "q" option, this allows a batch file to silently query if an application is up to date. Combined with the "n" option, a batch file could be produced containing the commands necessary to update the application. ═══ 8.1.3.8. i ═══ ignore return status of all commands executed The "i" option is equivalent to the .IGNORE directive. ═══ 8.1.3.9. k ═══ on error/interrupt: continue on next target Make will stop updating targets when a non-zero status is returned by a command. The "k" option will continue processing targets that do not depend on the target that caused the error. The .CONTINUE directive in a makefile will enable this error handling capability. ═══ 8.1.3.10. l ═══ the next parameter is the name of a output log file Make will output an error message when a non-zero status is returned by a command. The "l" option specifies a file that will record all error messages output by Make during the processing of the makefile. ═══ 8.1.3.11. m ═══ do not search for the MAKEINIT file The default action for Make is to search for an initialization file called "MAKEINIT". The "m" option will indicate to Make that processing of the MAKEINIT file is not desired. ═══ 8.1.3.12. ms ═══ Microsoft NMAKE mode The default action for Make is to process makefiles using Watcom syntax rules. The "ms" option will indicate to Make that it should process makefiles using Microsoft syntax rules. For example, the line continuation in NMAKE is a backslash ("\") at the end of the line. ═══ 8.1.3.13. n ═══ no execute mode - print commands without executing The "n" option will print out what commands should be executed to update the application without actually executing them. Combined with the "h" option, a batch file could be produced which would contain the commands necessary to update the application. Example: wmake /h /n >update.bat update This is useful for applications which require all available resources (memory and devices) for executing the updating commands. ═══ 8.1.3.14. o ═══ use circular implicit rule path When this option is specified, Make will use a circular path specification search which may save on disk activity for large makefiles. The "o" option is equivalent to the .OPTIMIZE directive. ═══ 8.1.3.15. p ═══ print out makefile information The "p" option will cause Make to print out information about all the explicit rules, implicit rules, and macro definitions. ═══ 8.1.3.16. q ═══ query mode - check targets without updating them The "q" option will cause Make to return a status of 1 if the application requires updating; it will return a status of 0 otherwise. Here is a example batch file using the "q" option: Example: wmake /q if errorstatus 0 goto noupdate wmake /q /h /n >\tmp\update.bat call \tmp\update.bat :noupdate ═══ 8.1.3.17. r ═══ do not use default definitions The default definitions are: __MAKEOPTS__ = __MAKEFILES__ = __VERSION__ = __LOADDLL__= defined if DLL loading supported __MSDOS__ = defined if MS/DOS version __WINDOWS__ = defined if Windows version __NT__ = defined if Windows NT version __NT386__ = defined if 32-bit Windows NT version __OS2__ = defined if OS/2 version __QNX__ = defined if QNX version #endif # clear .EXTENSIONS list .EXTENSIONS: # In general, # set .EXTENSIONS list as follows .EXTENSIONS: .exe .nlm .dsk .lan .exp & .lib .obj & .i & .asm .c .cpp .cxx .cc .for .pas .cob & .h .hpp .hxx .hh .fi .mif .inc For Microsoft NMAKE compatibility (when you use the "ms" option), the following default definitions are established. # For Microsoft NMAKE compatibility switch, # set .EXTENSIONS list as follows .EXTENSIONS: .exe .obj .asm .c .cpp .cxx & .bas .cbl .for .f .f90 .pas .res .rc AS=ml BC=bc CC=cl COBOL=cobol CPP=cl CXX=cl FOR=fl PASCAL=pl RC=rc .asm.exe: $(AS) $(AFLAGS) $*.asm .asm.obj: $(AS) $(AFLAGS) /c $*.asm .c.exe: $(CC) $(CFLAGS) $*.c .c.obj: $(CC) $(CFLAGS) /c $*.c .cpp.exe: $(CPP) $(CPPFLAGS) $*.cpp .cpp.obj: $(CPP) $(CPPFLAGS) /c $*.cpp .cxx.exe: $(CXX) $(CXXFLAGS) $*.cxx .cxx.obj: $(CXX) $(CXXFLAGS) $*.cxx .bas.obj: $(BC) $(BFLAGS) $*.bas .cbl.exe: $(COBOL) $(COBFLAGS) $*.cbl, $*.exe; .cbl.obj: $(COBOL) $(COBFLAGS) $*.cbl; .f.exe: $(FOR) $(FFLAGS) $*.f .f.obj: $(FOR) /c $(FFLAGS) $*.f .f90.exe: $(FOR) $(FFLAGS) $*.f90 .f90.obj: $(FOR) /c $(FFLAGS) $*.f90 .for.exe: $(FOR) $(FFLAGS) $*.for .for.obj: $(FOR) /c $(FFLAGS) $*.for .pas.exe: $(PASCAL) $(PFLAGS) $*.pas .pas.obj: $(PASCAL) /c $(PFLAGS) $*.pas .rc.res: $(RC) $(RFLAGS) /r $* For OS/2, the __MSDOS__ macro will be replaced by __OS2__ and for Windows NT, the __MSDOS__ macro will be replaced by __NT__. The "r" option will disable these definitions before processing any makefiles. ═══ 8.1.3.18. s ═══ silent mode - do not print commands before execution The "s" option is equivalent to the .SILENT directive. ═══ 8.1.3.19. t ═══ touch files instead of executing commands Sometimes there are changes which are purely cosmetic (adding a comment to a source file) that will cause targets to be updated needlessly thus wasting computer resources. The "t" option will make files appear younger without altering their contents. The "t" option is useful but should be used with caution. ═══ 8.1.3.20. u ═══ UNIX compatibility mode The "u" option will indicate to Make that the line continuation character should be a backslash "\" rather than an ampersand "&". ═══ 8.1.3.21. z ═══ do not erase target after error/interrupt (disables prompting) The "z" option will indicate to Make that if an error or interrupt occurs during makefile processing then the current target being made should not be deleted. The .HOLD directive in a makefile has the same effect as the "z" option. ═══ 8.1.4. Special Macros ═══ Watcom Make has many different special macros. Here are some of the simpler ones. $$ represents the character "$" $# represents the character "#" $@ full file name of the target $* target with the extension removed $< list of all dependents $? list of dependents that are younger than the target The following macros are for more sophisticated makefiles. __MSDOS__ This macro is defined in the MS/DOS environment. __NT__ This macro is defined in the Windows NT environment. __OS2__ This macro is defined in the OS/2 environment. __MAKEOPTS__ contains all of the command line options that WMAKE was invoked with except for the "f" options __MAKEFILES__ contains the names of all of the makefiles processed at the time of expansion (includes the file currently being processed) The next three tables contain macros that are valid during execution of command lists for explicit rules, implicit rules, and the .ERROR directive. The expansion is presented for the following example: Example: a:\dir\target.ext : b:\dir1\dep1.ex1 c:\dir2\dep2.ex2 $^@ a:\dir\target.ext $^* a:\dir\target $^& target $^. target.ext $^: a:\dir\ $[@ b:\dir1\dep1.ex1 $[* b:\dir1\dep1 $[& dep1 $[. dep1.ex1 $[: b:\dir1\ $]@ c:\dir2\dep2.ex2 $]* c:\dir2\dep2 $]& dep2 $]. dep2.ex2 $]: c:\dir2\ ═══ 8.2. Dependency Declarations ═══ In order for Watcom Make to be effective, a list of file dependencies must be declared. The declarations may be entered into a text file of any name but Make will read a file called "MAKEFILE" by default if it is invoked as follows: Example: C>wmake If you want to use a file that is not called "MAKEFILE" then the command line option "f" will cause Make to read the specified file instead of the default "MAKEFILE". Example: C>wmake /f myfile We will now go through an example to illustrate how Make may be used for a simple application. Suppose we have an input file, a report file, and a report generator program then we may declare a dependency as follows: # # (a comment in a makefile starts with a "#") # simple dependency declaration # balance.lst : ledger.dat doreport Note that the dependency declaration starts at the beginning of a line while commands always have at least one blank or tab before them. This form of a dependency declaration is called an explicit rule. The file "BALANCE.LST" is called the target of the rule. The dependent of the rule is the file "LEDGER.DAT" while "DOREPORT" forms one line of the rule command list. The dependent is separated from the target by a colon. Hint: A good habit to develop is to always put spaces around the colon so that it will not be confused with drive specifications (e.g., a:). The explicit rule declaration indicates to Make that the program "DOREPORT" should be executed if "LEDGER.DAT" is younger than "BALANCE.LST" or if "BALANCE.LST" does not yet exist. In general, if the dependent file has a more recent modification date and time than the target file then Watcom Make will execute the specified command. ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── Note: The terminology employed here is used by S.I.Feldman of Bell Laboratories in Make - A Program for Maintaining Computer Programs. Confusion often arises from the use of the word "dependent". In this context, it means "a subordinate part". In the example, "LEDGER.DAT" is a subordinate part of the report "BALANCE.LST". ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── ═══ 8.3. Multiple Dependents ═══ Suppose that our report "BALANCE.LST" becomes out-of-date if any of the files "LEDGER.DAT", "SALES.DAT" or "PURCHASE.DAT" are modified. We may modify the dependency rule as follows: # # multiple dependents rule # balance.lst : ledger.dat sales.dat purchase.dat doreport This is an example of a rule with multiple dependents. In this situation, the program "DOREPORT" should be executed if any of "LEDGER.DAT", "SALES.DAT" or "PURCHASE.DAT" are younger than "BALANCE.LST" or if "BALANCE.LST" does not yet exist. In cases where there are multiple dependents, if any of the dependent files has a more recent modification date and time than the target file then Watcom Make will execute the specified command. ═══ 8.4. Multiple Targets ═══ Suppose that the "DOREPORT" program produces two reports. If both of these reports require updating as a result of modification to the dependent files, we could change the rule as follows: # # multiple targets and multiple dependents rule # balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport Suppose that you entered the command: wmake which causes Make to start processing the rules described in "MAKEFILE". In the case where multiple targets are listed in the makefile, Make will, by default, process only the first target it encounters. In the example, Make will check the date and time of "BALANCE.LST" against its dependents since this is the first target listed. To indicate that some other target should be processed, the target is specified as an argument to the Make command. Example: wmake summary.lst There are a number of interesting points to consider: 1. By default, Make will only check that the target file exists after the command ("DOREPORT" in this example) is executed. It does not check that the target's time-stamp shows it to be younger. If the target file does not exist after the command has been executed, an error is reported. 2. There is no guarantee that the command you have specified does update the target file. In other words, simply because you have stated a dependency does not mean that one exists. 3. Furthermore, it is not implied that other targets in our list will not be updated. In the case of our example, you can assume that we have designed the "doreport" command to update both targets. ═══ 8.5. Multiple Rules ═══ A makefile may consist of any number of rules. Note that the following: target1 target2 : dependent1 dependent2 dependent3 command list is equivalent to: target1 : dependent1 dependent2 dependent3 command list target2 : dependent1 dependent2 dependent3 command list Also, the rules may depend on the targets of other rules. # # rule 1: this rule uses rule 2 # balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport # # rule 2: used by rules 1 and 3 # sales.dat : canada.dat england.dat usa.dat dosales # # rule 3: this rule uses rule 2 # year.lst : ledger.dat sales.dat purchase.dat doyearly The dependents are checked to see if they are the targets of any other rules in the makefile in which case they are updated. This process of updating dependents that are targets in other rules continues until a rule is reached that has only simple dependents that are not targets of rules. At this point, if the target does not exist or if any of the dependents is younger than the target then the command list associated with the rule is executed. Hint: The term "updating", in this context, refers to the process of checking the time-stamps of dependents and running the specified command list whenever they are out-of-date. Whenever a dependent is the target of some other rule, the dependent must be brought up-to-date first. Stated another way, if "A" depends on "B" and "B" depends on "C" and "C" is younger than "B" then we must update "B" before we update "A". Make will check to ensure that the target exists after its associated command list is executed. The target existence checking may be disabled in two ways: 1. use the command line option "c" 2. use the .NOCHECK directive. The rule checking returns to the previous rule that had the target as a dependent. Upon returning to the rule, the command list is executed if the target does not exist or if any of the updated dependents are now younger than the target. If you were to type: wmake here are the steps that would occur with the previous makefile: update(balance.lst) (rule 1) update(ledger.dat) (not a target) update(sales.dat) (found rule 2) update(canada.dat) (not a target) update(england.dat) (not a target) update(usa.dat) (not a target) IF sales.dat does not exist OR any of (canada.dat,england.dat,usa.dat) is younger than sales.dat THEN execute "dosales" update(purchase.dat) (not a target) IF balance.lst does not exist OR any of (ledger.dat,sales.dat,purchase.dat) is younger than (balance.lst) THEN execute "doreport" The third rule in the makefile will not be included in this update sequence of steps. Recall that the default target that is "updated" is the first target in the first rule encountered in the makefile. This is the default action taken by Make when no target is specified on the command line. If you were to type: wmake year.lst then the file "YEAR.LST" would be updated. As Make reads the rules in "MAKEFILE", it discovers that updating "YEAR.LST" involves updating "SALES.DAT". The update sequence is similar to the previous example. ═══ 8.6. Automatic Dependency Detection (.AUTODEPEND) ═══ Explicit listing of dependencies in a makefile can often be tedious in the development and maintenance phases of a project. The Watcom C/C++ compiler will insert dependency information into the object file as it processes source files so that a complete snapshot of the files necessary to build the object file are recorded. Since all files do not have dependency information contained within them in a standard form, it is necessary to indicate to Make when dependencies are present. To illustrate the use of the .AUTODEPEND directive, we will show its use in an implicit rule and in an explicit rule. # # .AUTODEPEND example # .c.obj: .AUTODEPEND wcc386 $[* $(compile_options) test.exe : a.obj b.obj c.obj test.res wlink FILE a.obj, b.obj, c.obj wrc /q /bt=windows test.res test.exe test.res : test.rc test.ico .AUTODEPEND wrc /ad /q /bt=windows /r $[@ $^@ In the above example, Make will use the contents of the object file to determine whether the object file has to be built during processing. The Watcom Resource Compiler can also insert dependency information into a resource file that can be used by Make. ═══ 8.7. Targets Without Any Dependents (.SYMBOLIC) ═══ There must always be at least one target in a rule but it is not necessary to have any dependents. If a target does not have any dependents, the command list associated with the rule will always be executed if the target is updated. You might ask, "What may a rule with no dependents be used for?". A rule with no dependents may be used to describe actions that are useful for the group of files being maintained. Possible uses include backing up files, cleaning up files, or printing files. To illustrate the use of the .SYMBOLIC directive, we will add two new rules to the previous example. First, we will omit the .SYMBOLIC directive and observe what will happen when it is not present. # # rule 4: backup the data files # backup : echo "insert backup disk" pause copy *.dat a: echo "backup complete" # # rule 5: cleanup temporary files # cleanup : del *.tmp del \tmp\*.* and then execute the command: wmake backup Make will execute the command list associated with the "backup" target and issue an error message indicating that the file "BACKUP" does not exist after the command list was executed. The same thing would happen if we typed: wmake cleanup In this makefile we are using "backup" and "cleanup" to represent actions we want performed. The names are not real files but rather they are symbolic names. This special type of target may be declared with the .SYMBOLIC directive. This time, we show rules 4 and 5 with the appropriate addition of .SYMBOLIC directives. # # rule 4: backup the data files # backup : .SYMBOLIC echo "insert backup disk" pause copy *.dat a: echo "backup complete" # # rule 5: cleanup temporary files # cleanup : .SYMBOLIC del *.tmp del \tmp\*.* The use of the .SYMBOLIC directive indicates to Make that the target should always be updated internally after the command list associated with the rule has been executed. A short form for the common idiom of singular .SYMBOLIC targets like: target : .SYMBOLIC commands is: target commands This kind of target definition is useful for many types of management tasks that can be described in a makefile. ═══ 8.8. Preserving Targets (.PRECIOUS) ═══ Most operating system utilities and programs have special return codes that indicate error conditions. Watcom Make will check the return code for every command executed. If the return code is non-zero, Make will stop processing the current rule and optionally delete the current target being updated. If a file is precious enough that this treatment of return codes is not wanted then the .PRECIOUS directive may be used. The .PRECIOUS directive indicates to Make that the target should not be deleted if an error occurs during the execution of the associated command list. Here is an example of the .PRECIOUS directive: # # .PRECIOUS example # balance.lst summary.lst : ledger.dat sales.dat purchase.dat .PRECIOUS doreport If the program "DOREPORT" executes and its return code is non-zero then Make will not attempt to delete "BALANCE.LST" or "SUMMARY.LST". If only one of the files is precious then the makefile could be coded as follows: # # .PRECIOUS example # balance.lst : .PRECIOUS balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport The file "BALANCE.LST" will not be deleted if an error occurs while the program "DOREPORT" is executing. ═══ 8.9. Ignoring Return Codes (.IGNORE) ═══ Some programs do not have meaningful return codes so for these programs we want to ignore the return code completely. There are different ways to ignore return codes namely, 1. use the command line option "i" 2. put a "-" in front of specific commands, or 3. use the .IGNORE directive. In the following example, the rule: # # ignore return code example # balance.lst summary.lst : ledger.dat sales.dat purchase.dat -doreport will ignore the return status from the program "DOREPORT". Using the dash in front of the command is the preferred method for ignoring return codes because it allows Make to check all the other return codes. The .IGNORE directive is used as follows: # # .IGNORE example # .IGNORE balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport Using the .IGNORE directive will cause Make to ignore the return code for every command. The "i" command line option and the .IGNORE directive prohibit Make from performing any error checking on the commands executed and, as such, should be used with caution. Another way to handle non-zero return codes is to continue processing targets which do not depend on the target that had a non-zero return code during execution of its associated command list. There are two ways of indicating to Make that processing should continue after a non-zero return code: 1. use the command line option "k" 2. use the .CONTINUE directive. ═══ 8.10. Erasing Targets After Error (.ERASE) ═══ Most operating system utilities and programs have special return codes that indicate error conditions. Watcom Make will check the return code for every command executed. If the return code is non-zero, Make will stop processing the current rule and optionally delete the current target being updated. By default, Make will prompt for deletion of the current target. The .ERASE directive indicates to Make that the target should be deleted if an error occurs during the execution of the associated command list. No prompt is issued in this case. Here is an example of the .ERASE directive: # # .ERASE example # .ERASE balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport If the program "DOREPORT" executes and its return code is non-zero then Make will attempt to delete "BALANCE.LST" or "SUMMARY.LST" depending on which it was updating. ═══ 8.11. Preserving Targets After Error (.HOLD) ═══ Most operating system utilities and programs have special return codes that indicate error conditions. Watcom Make will check the return code for every command executed. If the return code is non-zero, Make will stop processing the current rule and optionally delete the current target being updated. By default, Make will prompt for deletion of the current target. The .HOLD directive indicates to Make that the target should not be deleted if an error occurs during the execution of the associated command list. No prompt is issued in this case. The .HOLD directive is similar to .PRECIOUS but applies to all targets listed in the makefile. Here is an example of the .HOLD directive: # # .HOLD example # .HOLD balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport If the program "DOREPORT" executes and its return code is non-zero then Make will not delete "BALANCE.LST" or "SUMMARY.LST". ═══ 8.12. Suppressing Terminal Output (.SILENT) ═══ As commands are executed, Watcom Make will print out the current command before it is executed. It is possible to execute the makefile without having the commands printed. There are three ways to inhibit the printing of the commands before they are executed, namely: 1. use the command line option "s" 2. put an "@" in front of specific commands, or 3. use the .SILENT directive. In the following example, the rule: # # silent command example # balance.lst summary.lst : ledger.dat sales.dat purchase.dat @doreport will prevent the string "doreport" from being printed on the screen before the command is executed. The .SILENT directive is used as follows: # # .SILENT example # .SILENT balance.lst summary.lst : ledger.dat sales.dat purchase.dat doreport Using the .SILENT directive or the "s" command line option will inhibit the printing of all commands before they are executed. At this point, most of the capability of Make may be realized. Methods for making makefiles more succinct will be discussed. ═══ 8.13. Macros ═══ Watcom Make has a simple macro facility that may be used to improve makefiles by making them easier to read and maintain. A macro identifier may be composed from a string of alphabetic characters and numeric characters. The underscore character is also allowed in a macro identifier. If the macro identifier starts with a "%" character, the macro identifier represents an environment variable. For instance, the macro identifier "%path" represents the environment variable "path". 2morrow yes stitch_in_9 yes invalid~id no 2b_or_not_2b yes %path yes reports yes !@#*% no We will use a programming example to show how macros are used. The programming example involves four C/C++ source files and two header files. Here is the initial makefile (before macros): # # programming example # (before macros) # plot.exe : main.obj input.obj calc.obj output.obj wlink @plot main.obj : main.c defs.h globals.h wcc386 main /mf /d1 /w3 calc.obj : calc.c defs.h globals.h wcc386 calc /mf /d1 /w3 input.obj : input.c defs.h globals.h wcc386 input /mf /d1 /w3 output.obj : output.c defs.h globals.h wcc386 output /mf /d1 /w3 Macros become useful when changes must be made to makefiles. If the programmer wanted to change the compiler options for the different compiles, the programmer would have to make a global change to the makefile. With this simple example, it is quite easy to make the change but try to imagine a more complex example with different programs having similar options. The global change made by the editor could cause problems by changing the options for other programs. A good habit to develop is to define macros for any programs that have command line options. In our example, we would change the makefile to be: # # programming example # (after macros) # link_options = compiler = wcc386 compile_options = /mf /d1 /w3 plot.exe : main.obj input.obj calc.obj output.obj wlink $(link_options) @plot main.obj : main.c defs.h globals.h $(compiler) main $(compile_options) calc.obj : calc.c defs.h globals.h $(compiler) calc $(compile_options) input.obj : input.c defs.h globals.h $(compiler) input $(compile_options) output.obj : output.c defs.h globals.h $(compiler) output $(compile_options) A macro definition consists of a macro identifier starting on the beginning of the line followed by an "=" which in turn is followed by the text to be replaced. A macro may be redefined, with the latest declaration being used for subsequent expansions (no warning is given upon redefinition of a macro). The replacement text may contain macro references. A macro reference may occur in two forms. The previous example illustrates one way to reference macros whereby the macro identifier is delimited by "$(" and ")". The parentheses are optional so the macros "compiler" and "compile_options" could be referenced by: main.obj : main.c defs.h globals.h $compiler main $compile_options Certain ambiguities may arise with this form of macro reference. For instance, examine this makefile fragment: Example: temporary_dir = \tmp\ temporary_file = $temporary_dirtmp000.tmp The intention of the declarations is to have a macro that will expand into a file specification for a temporary file. Make will collect the largest identifier possible before macro expansion occurs. The macro reference is followed by text that looks like part of the macro identifier ("tmp000") so the macro identifier that will be referenced will be "temporary_dirtmp000". The incorrect macro identifier will not be defined so an error message will be issued. If the makefile fragment was: temporary_dir = \tmp\ temporary_file = $(temporary_dir)tmp000.tmp there would be no ambiguity. The preferred way to reference macros is to enclose the macro identifier by "$(" and ")". Macro references are expanded immediately on dependency lines (and thus may not contain references to macros that have not been defined) but other macro references have their expansion deferred until they are used in a command. In the previous example, the macros "link_options", "compiler", and "compile_options" will not be expanded until the commands that reference them are executed. Another use for macros is to replace large amounts of text with a much smaller macro reference. In our example, we only have two header files but suppose we had very many header files. Each explicit rule would be very large and difficult to read and maintain. We will use the previous example makefile to illustrate this use of macros. # # programming example # (with more macros) # link_options = compiler = wcc386 compile_options = /mf /d1 /w3 header_files = defs.h globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) wlink $(link_options) @plot main.obj : main.c $(header_files) $(compiler) main $(compile_options) calc.obj : calc.c $(header_files) $(compiler) calc $(compile_options) input.obj : input.c $(header_files) $(compiler) input $(compile_options) output.obj : output.c $(header_files) $(compiler) output $(compile_options) Notice the ampersand ("&") at the end of the macro definition for "object_files". The ampersand indicates that the macro definition continues on the next line. In general, if you want to continue a line in a makefile, use an ampersand ("&") at the end of the line. There are special macros provided by Make to access environment variable names. To access the PATH environment variable in a makefile, we use the macro identifier "%path". For example, if we have the following line in a command list: Example: echo $(%path) it will print out the current value of the PATH environment variable when it is executed. There are two other special environment macros that are predefined by Make. The macro identifier "%cdrive" will expand into one letter representing the current drive. The macro identifier "%cwd" will expand into the current working directory. These macro identifiers are not very useful unless we can specify that they be expanded immediately. The complementary macros "$+" and "$-" respectively turn on and turn off immediate expansion of macros. The scope of the "$+" macro is the current line after which the default macro expansion behaviour is resumed. A possible use of these macros is illustrated by the following example makefile. # # $(%cdrive), $(%cwd), $+, and $- example # dir1 = $(%cdrive):$(%cwd) dir2 = $+ $(dir1) $- example : .SYMBOLIC cd .. echo $(dir1) echo $(dir2) Which would produce the following output if the current working directory is C:\WATCOM\SOURCE\EXAMPLE: Example: (command output only) C:\WATCOM\SOURCE C:\WATCOM\SOURCE\EXAMPLE The macro definition for "dir2" forces immediate expansion of the "%cdrive" and "%cwd" macros thus defining "dir2" to be the current directory that Make was invoked in. The macro "dir1" is not expanded until execution time when the current directory has changed from the initial directory. Combining the $+ and $- special macros with the special macro identifiers "%cdrive" and "%cwd" is a useful makefile technique. The $+ and $- special macros are general enough to be used in many different ways. Constructing other macros is another use for the $+ and $- special macros. Make allows macros to be redefined and combining this with the $+ and $- special macros, similar looking macros may be constructed. # # macro construction with $+ and $- # template = file1.$(ext) file2.$(ext) file3.$(ext) file4.$(ext) ext = dat data_files = $+ $(template) $- ext = lst listing_files = $+ $(template) $- example : .SYMBOLIC echo $(data_files) echo $(listing_files) This makefile would produce the following output: Example: file1.dat file2.dat file3.dat file4.dat file1.lst file2.lst file3.lst file4.lst Adding more text to a macro can also be done with the $+ and $- special macros. # # macro addition with $+ and $- # objs = file1.obj file2.obj file3.obj objs = $+$(objs)$- file4.obj objs = $+$(objs)$- file5.obj example : .SYMBOLIC echo $(objs) This makefile would produce the following output: Example: file1.obj file2.obj file3.obj file4.obj file5.obj Make provides a shorthand notation for this type of macro operation. Text can be added to a macro by using the "+=" macro assignment. The previous makefile can be written as: # # macro addition with += # objs = file1.obj file2.obj file3.obj objs += file4.obj objs += file5.obj example : .SYMBOLIC echo $(objs) and still produce the same results. The shorthand notation "+=" supported by Make provides a quick way to add more text to macros. There are instances when it is useful to have macro identifiers that have macro references contained in them. If you wanted to print out an informative message before linking the executable that was different between the debugging and production version, we would express it as follows: # # programming example # (macro selection) # version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = defs.h globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @plot main.obj : main.c $(header_files) $(compiler) main $(compile_options) calc.obj : calc.c $(header_files) $(compiler) calc $(compile_options) input.obj : input.c $(header_files) $(compiler) input $(compile_options) output.obj : output.c $(header_files) $(compiler) output $(compile_options) Take notice of the macro references that are of the form: $($(version)) The expansion of a macro reference begins by expanding any macros seen until a matching right parenthesis is found. The macro identifier that is present after the matching parenthesis is found will be expanded. The other form of macro reference namely: $ may be used in a similar fashion. The previous example would be of the form: $$version Macro expansion occurs until a character that cannot be in a macro identifier is found (on the same line as the "$") after which the resultant macro identifier is expanded. If you want two macros to be concatenated then the line would have to be coded: $(macro1)$(macro2) The use of parentheses is the preferred method for macro references because it completely specifies the order of expansion. In the previous example, we can see that the four command lines that invoke the compiler are very similar in form. We may make use of these similarities by denoting the command by a macro reference. We need to be able to define a macro that will expand into the correct command when processed. Fortunately, Make can reference the first member of the dependent list, the last member of the dependent list, and the current target being updated with the use of some special macros. These special macros have the form: $ where is one of: "^" represents the current target being updated "[" represents the first member of the dependent list "]" represents the last member of the dependent list and is one of: "@" full file name "*" file name with extension removed "&" file name with path and extension removed "." file name with path removed ":" path of file name If the file "D:\DIR1\DIR2\NAME.EXT" is the current target being updated then the following example will show how the form qualifiers are used. $^@ D:\DIR1\DIR2\NAME.EXT $^* D:\DIR1\DIR2\NAME $^& NAME $^. NAME.EXT $^: D:\DIR1\DIR2\ These special macros provide the capability to reference targets and dependents in a variety of ways. # # programming example # (more macros) # version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) compiler_command = wcc386 $[* $(compile_options) header_files = defs.h globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* main.obj : main.c $(header_files) $(compiler_command) calc.obj : calc.c $(header_files) $(compiler_command) input.obj : input.c $(header_files) $(compiler_command) output.obj : output.c $(header_files) $(compiler_command) This example illustrates the use of the special dependency macros. Notice the use of "$^*" in the linker command. The macro expands into the string "plot" since "plot.exe" is the target when the command is processed. The use of the special dependency macros is recommended because they make use of information that is already contained in the dependency rule. At this point, we know that macro references begin with a "$" and that comments begin with a "#". What happens if we want to use these characters without their special meaning? Make has two special macros that provide these characters to you. The special macro "$$" will result in a "$" when expanded and "$#" will expand into a "#". These special macros are provided so that you are not forced to work around the special meanings of the "$" and "#" characters. ═══ 8.14. Implicit Rules ═══ Watcom Make is capable of accepting declarations of commonly used dependencies. These declarations are called "implicit rules" as opposed to "explicit rules" which were discussed previously. Implicit rules may be applied only in instances where you are able to describe a dependency in terms of file extensions. Hint: Recall that a file extension is the portion of the file name which follows the period. In the file specification: C:\DOS\ANSI.SYS the file extension is "SYS". An implicit rule provides a command list for a dependency between files with certain extensions. The form of an implicit rule is as follows: ..: Implicit rules are used if a file has not been declared as a target in any explicit rule or the file has been declared as a target in an explicit rule with no command list. For a given target file, a search is conducted to see if there are any implicit rules defined for the target file's extension in which case Make will then check if the file with the dependent extension in the implicit rule exists. If the file with the dependent extension exists then the command list associated with the implicit rule is executed and processing of the makefile continues. Other implicit rules for the target extension are searched in a similar fashion. The order in which the dependent extensions are checked becomes important if there is more than one implicit rule declaration for a target extension. If we have the following makefile fragment: Example: .pas.obj: (command list) .c.obj: (command list) an ambiguity arises. If we have a target file "TEST.OBJ" then which do we check for first, "TEST.PAS" or "TEST.C"? Make handles this with the .EXTENSIONS directive. The .EXTENSIONS directive declares which extensions are allowed to be used in implicit rules and how these extensions are ordered. The default .EXTENSIONS declaration is: .EXTENSIONS: .EXTENSIONS: .exe .exp .lib .obj .asm .c .for .pas .cob .h .fi .mif A .EXTENSIONS directive with an empty list will clear the .EXTENSIONS list and any previously defined implicit rules. Any subsequent .EXTENSIONS directives will add extensions to the end of the list. Hint: The default .EXTENSIONS declaration could have been coded as: .EXTENSIONS: .EXTENSIONS: .exe .EXTENSIONS: .exp .EXTENSIONS: .lib .EXTENSIONS: .obj .EXTENSIONS: .asm .c .for .pas .cob .h .fi .mif with identical results. Make will not allow any implicit rule declarations that use extensions that are not in the current .EXTENSIONS list. Returning to our makefile fragment: .pas.obj: (command list) .c.obj: (command list) and our target file "TEST.OBJ", we now know that the .EXTENSIONS list determines in what order the dependents "TEST.PAS" and "TEST.C" will be tried. If the .EXTENSIONS declaration is: Example: .EXTENSIONS: .EXTENSIONS: .exe .obj .asm .pas .c .cpp .for .cob we can see that the dependent file "TEST.PAS" will be tried first as a possible dependent with "TEST.C" being tried next. One apparent problem with implicit rules and their associated command lists is that they are used for many different targets and dependents during the processing of a makefile. The same problem occurs with commands constructed from macros. Recall that there is a set of special macros that start with "$^", "$[", or "$]" that reference the target, first dependent, or last dependent of an explicit dependency rule. In an implicit rule there may be only one dependent or many dependents depending on whether the rule is being executed for a target with a single colon ":" or double colon "::" dependency. If the target has a single colon or double colon dependency, the "$^", "$[", and "$]" special macros will reflect the values in the rule that caused the implicit rule to be invoked. Otherwise, if the target does not have a dependency rule then the "$[" and "$]" special macros will be set to the same value, namely, the file found in the implicit rule search. We will use the last programming example to illustrate a possible use of implicit rules. # # programming example # (implicit rules) # version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = defs.h globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c.obj: $(compiler) $[* $(compile_options) main.obj : main.c $(header_files) calc.obj : calc.c $(header_files) input.obj : input.c $(header_files) output.obj : output.c $(header_files) As this makefile is processed, any time an object file is found to be older than its associated source file or header files then Make will attempt to execute the command list associated with the explicit rule. Since there are no command lists associated with the four object file targets, an implicit rule search is conducted. Suppose "CALC.OBJ" was older than "CALC.C". The lack of a command list in the explicit rule with "CALC.OBJ" as a target causes the ".c.obj" implicit rule to be invoked for "CALC.OBJ". The file "CALC.C" is found to exist so the commands wcc386 calc /mf /d1 /w3 echo linking debug version ... wlink debug all @plot are executed. The last two commands are a result of the compilation of "CALC.C" producing a "CALC.OBJ" file that is younger than the "PLOT.EXE" file that in turn must be generated again. The use of implicit rules is straightforward when all the files that the makefile deals with are in the current directory. Larger applications may have files that are in many different directories. Suppose we moved the programming example files to three sub-directories. *.H \EXAMPLE\H *.C \EXAMPLE\C rest \EXAMPLE\O Now the previous makefile (located in the \EXAMPLE\O sub-directory) would look like this: # # programming example # (implicit rules) # h_dir = \example\h\ #sub-directory containing header files c_dir = \example\c\ #sub-directory containing C/C++ files version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = $(h_dir)defs.h $(h_dir)globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c.obj: $(compiler) $[* $(compile_options) main.obj : $(c_dir)main.c $(header_files) calc.obj : $(c_dir)calc.c $(header_files) input.obj : $(c_dir)input.c $(header_files) output.obj : $(c_dir)output.c $(header_files) Suppose "\EXAMPLE\O\CALC.OBJ" was older than "\EXAMPLE\C\CALC.C". The lack of a command list in the explicit rule with "CALC.OBJ" as a target causes the ".c.obj" implicit rule to be invoked for "CALC.OBJ". At this time, the file "\EXAMPLE\O\CALC.C" is not found so an error is reported indicating that "CALC.OBJ" could not be updated. How may implicit rules be useful in larger applications if they will only search the current directory for the dependent file? We must specify more information about the dependent extension (in this case ".C"). We do this by associating a path with the dependent extension as follows: . : This allows the implicit rule search to find the files with the dependent extension. Hint: A valid path specification is made up of directory specifications separated by semicolons (";"). Here are some path specifications: D:;C:\DOS;C:\UTILS;C:\WC C:\SYS A:\BIN;D: Notice that these path specifications are identical to the form required by the operating system shell's "PATH" command. Our makefile will be correct now if we add the new declaration as follows: # # programming example # (implicit rules) # h_dir = \example\h\ #sub-directory containing header files c_dir = \example\c\ #sub-directory containing C/C++ files version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = $(h_dir)defs.h $(h_dir)globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c: $(c_dir) .c.obj: $(compiler) $[* $(compile_options) main.obj : $(c_dir)main.c $(header_files) calc.obj : $(c_dir)calc.c $(header_files) input.obj : $(c_dir)input.c $(header_files) output.obj : $(c_dir)output.c $(header_files) Suppose "\EXAMPLE\O\CALC.OBJ" is older than "\EXAMPLE\C\CALC.C". The lack of a command list in the explicit rule with "CALC.OBJ" as a target will cause the ".c.obj" implicit rule to be invoked for "CALC.OBJ". The dependent extension ".C" has a path associated with it so the file "\EXAMPLE\C\CALC.C" is found to exist. The commands wcc386 \EXAMPLE\C\CALC /mf /d1 /w3 echo linking debug version ... wlink debug all @plot are executed to update the necessary files. If the application requires many source files in different directories Make will search for the files using their associated path specifications. For instance, if the current example files were setup as follows: \EXAMPLE\H DEFS.H, GLOBALS.H \EXAMPLE\C\PROGRAM MAIN.C, CALC.C \EXAMPLE\C\SCREEN INPUT.C, OUTPUT.C \EXAMPLE\O PLOT.EXE, MAKEFILE, MAIN.OBJ, CALC.OBJ, INPUT.OBJ, OUTPUT.OBJ the makefile would be changed to: # # programming example # (implicit rules) # h_dir = ..\h\ # sub-directory with header files # sub-directories with C/C++ source files program_dir = ..\c\program\ # - MAIN.C, CALC.C screen_dir = ..\c\screen\ # - INPUT.C, OUTPUT.C version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = $(h_dir)defs.h $(h_dir)globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c: $(program_dir);$(screen_dir) .c.obj: $(compiler) $[* $(compile_options) main.obj : $(program_dir)main.c $(header_files) calc.obj : $(program_dir)calc.c $(header_files) input.obj : $(screen_dir)input.c $(header_files) output.obj : $(screen_dir)output.c $(header_files) Suppose that there is a change in the "DEFS.H" file which causes all the source files to be recompiled. The implicit rule ".c.obj" is invoked for every object file so the corresponding ".C" file must be found for each ".OBJ" file. We will show where Make searches for the C/C++ source files. update main.obj test ..\c\program\main.c (it does exist) execute wcc386 ..\c\program\main /mf /d1 /w3 update calc.obj test ..\c\program\calc.c (it does exist) execute wcc386 ..\c\program\calc /mf /d1 /w3 update input.obj test ..\c\program\input.c (it does not exist) test ..\c\screen\input.c (it does exist) execute wcc386 ..\c\screen\input /mf /d1 /w3 update output.obj test ..\c\program\output.c (it does not exist) test ..\c\screen\output.c (it does exist) execute wcc386 ..\c\screen\output /mf /d1 /w3 etc. Notice that Make checked the sub-directory "..\C\PRORGAM" for the files "INPUT.C" and "OUTPUT.C". Make optionally may use a circular path specification search which may save on disk activity for large makefiles. The circular path searching may be used in two different ways: 1. use the command line option "o" 2. use the .OPTIMIZE directive. Make will retain (for each suffix) what sub-directory yielded the last successful search for a file. The search for a file is resumed at this directory in the hope that wasted disk activity will be minimized. If the file cannot be found in the sub-directory then Make will search the next sub-directory in the path specification (cycling to the first sub-directory in the path specification after an unsuccessful search in the last sub-directory). Changing the previous example to include this feature, results in the following: # # programming example # (optimized path searching) # .OPTIMIZE h_dir = ..\h\ # sub-directory with header files # sub-directories with C/C++ source files program_dir = ..\c\program\ # - MAIN.C, CALC.C screen_dir = ..\c\screen\ # - INPUT.C, OUTPUT.C version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) header_files = $(h_dir)defs.h $(h_dir)globals.h object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c: $(program_dir);$(screen_dir) .c.obj: $(compiler) $[* $(compile_options) main.obj : $(program_dir)main.c $(header_files) calc.obj : $(program_dir)calc.c $(header_files) input.obj : $(screen_dir)input.c $(header_files) output.obj : $(screen_dir)output.c $(header_files) Suppose again that there is a change in the "DEFS.H" file which causes all the source files to be recompiled. We will show where Make searches for the C/C++ source files using the optimized path specification searching. update main.obj test ..\c\program\main.c (it does exist) execute wcc386 ..\c\program\main /mf /d1 /w3 update calc.obj test ..\c\program\calc.c (it does exist) execute wcc386 ..\c\program\calc /mf /d1 /w3 update input.obj test ..\c\program\input.c (it does not exist) test ..\c\screen\input.c (it does exist) execute wcc386 ..\c\screen\input /mf /d1 /w3 update output.obj test ..\c\screen\output.c (it does exist) execute wcc386 ..\c\screen\output /mf /d1 /w3 etc. Make did not check the sub-directory "..\C\PROGRAM" for the file "OUTPUT.C" because the last successful attempt to find a ".C" file occurred in the "..\C\SCREEN" sub-directory. In this small example, the amount of disk activity saved by Make is not substantial but the savings become much more pronounced in larger makefiles. Hint: The simple heuristic method that Make uses for optimizing path specification searches namely, keeping track of the last successful sub-directory, is very effective in reducing the amount of disk activity during the processing of a makefile. A pitfall to avoid is having two files with the same name in the path. The version of the file that is used to update the target depends on the previous searches. Care should be taken when using files that have the same name with path specifications. Large makefiles for projects written in C/C++ may become difficult to maintain with all the header file dependencies. Ignoring header file dependencies and using implicit rules may reduce the size of the makefile while keeping most of the functionality intact. The previous example may be made smaller by using this idea. # # programming example # (no header dependencies) # .OPTIMIZE h_dir = ..\h\ # sub-directory with header files # sub-directories with C/C++ source files program_dir = ..\c\program\ # - MAIN.C, CALC.C screen_dir = ..\c\screen\ # - INPUT.C, OUTPUT.C version = debugging # debugging version msg_production = linking production version ... msg_debugging = linking debug version ... link_options_production = link_options_debugging = debug all link_options = $(link_options_$(version)) compiler = wcc386 compile_options_production = /mf /w3 compile_options_debugging = /mf /d1 /w3 compile_options = $(compile_options_$(version)) object_files = main.obj input.obj calc.obj & output.obj plot.exe : $(object_files) echo $(msg_$(version)) wlink $(link_options) @$^* .c: $(program_dir);$(screen_dir) .c.obj: $(compiler) $[* $(compile_options) Implicit rules are very useful in this regard providing you are aware that you have to make up for the information that is missing from the makefile. In the case of C/C++ programs, you must ensure that you force Make to compile any programs affected by changes in header files. Forcing Make to compile programs may be done by touching source files (not recommended), deleting object files, or using the "a" option and targets on the command line. Here is how the files "INPUT.OBJ" and "MAIN.OBJ" may be recompiled if a change in some header file affects both files. Example: del input.obj del main.obj wmake or using the "a" option Example: wmake /a input.obj main.obj The possibility of introducing bugs into programs is present when using this makefile technique because it does not protect the programmer completely from object modules becoming out-of-date. The use of implicit rules without header file dependencies is a viable makefile technique but it is not without its pitfalls. ═══ 8.15. Double Colon Explicit Rules ═══ Single colon ":" explicit rules are useful in many makefile applications. However, the single colon rule has certain restrictions that make it difficult to express more complex dependency relationships. The restrictions imposed on single colon ":" explicit rules are: 1. only one command list is allowed for each target 2. after the command list is executed, the target is considered up to date The first restriction becomes evident when you want to update a target in different ways (i.e., when the target is out of date with respect to different dependents). The double colon explicit rule removes this restriction. # # multiple command lists # target1 :: dependent1 dependent2 command1 target1 :: dependent3 dependent4 command2 Notice that if "target1" is out of date with respect to either "dependent1" or "dependent2" then "command1" will be executed. The double colon "::" explicit rule does not consider the target (in this case "target1") up to date after the command list is executed. Make will continue to attempt to update "target1". Afterwards "command2" will be executed if "target1" is out of date with respect to either "dependent3" or "dependent4". It is possible that both "command1" and "command2" will be executed. As a result of the target not being considered up to date, an implicit rule search will be conducted on "target1" also. Make will process the double colon "::" explicit rules in the order that they are encountered in the makefile. A useful application of the double colon "::" explicit rule involves maintaining and using prototype information generated by a compiler. # # double colon "::" example # compiler = wcc386 options = /w3 # generate macros for the .OBJ and .DEF files template = module1.$(ext) module2.$(ext) module3.$(ext) ext = obj objs = $+ $(template) $- ext = def defs = $+ $(template) $- # add .DEF to the extensions list .EXTENSIONS: .EXTENSIONS: .exe .obj .def .c # implicit rules for the .OBJ and .DEF files .c.obj: $(compiler) $[* $(options) # generate the prototype file (only do a syntax check) .c.def: $(compiler) $[* $(options) /v/zs program.exe :: $(defs) erase *.err program.exe :: $(objs) wlink @$^* The ".OBJ" files are updated to complete the update of the file "PROGRAM.EXE". It is important to keep in mind that Make does not consider the file "PROGRAM.EXE" up to date until it has conducted a final implicit rule search. The double colon "::" explicit rule is useful when describing complex update actions. ═══ 8.16. Preprocessing Directives ═══ One of the primary objectives in using a make utility is to improve the development and maintenance of projects. A programming project consisting of many makefiles in different sub-directories may become unwieldy to maintain. The maintenance problem stems from the amount of duplicated information scattered throughout the project makefiles. Make provides a method to reduce the amount of duplicated information present in makefiles. Preprocessing directives provide the capability for different makefiles to make use of common information. ═══ 8.16.1. File Inclusion ═══ A common solution to the "duplicated information" problem involves referencing text contained in one file from many different files. Make supports file inclusion with the !include preprocessing directive. The development of object libraries, using 16-bit Watcom C/C++, for the different 80x86 16-bit memory models provides an ideal example to illustrate the use of the !include preprocessing directive. \WINDOW WINDOW.CMD, WINDOW.MIF \WINDOW\H PROTO.H, GLOBALS.H, BIOS_DEF.H \WINDOW\C WINDOW.C, KEYBOARD.C, MOUSE.C, BIOS.C \WINDOW\SCSD small model object files, MAKEFILE, WINDOW_S.LIB \WINDOW\SCBD compact model object files, MAKEFILE, WINDOW_C.LIB \WINDOW\BCSD medium model object files, MAKEFILE, WINDOW_M.LIB \WINDOW\BCBD large model object files, MAKEFILE, WINDOW_L.LIB \WINDOW\BCHD huge model object files, MAKEFILE, WINDOW_L.LIB The WLIB command file "WINDOW.CMD" contains the list of library operations required to build the libraries. The contents of "WINDOW.CMD" are: -+window -+bios -+keyboard -+mouse The "-+" library manager command indicates to WLIB that the object file should be replaced in the library. The file "WINDOW.MIF" contains the makefile declarations that are common to every memory model. The ".MIF" extension will be used for all the Make Include Files discussed in this manual. This extension is also in the default extension list so it is a recommended extension for Make include files. The contents of the "WINDOW.MIF" file is as follows: # # example of a Make Include File # common = /d1 /w3 # common options objs = window.obj bios.obj keyboard.obj mouse.obj .c: ..\c .c.obj: wcc $[* $(common) $(local) /m$(model) window_$(model).lib : $(objs) wlib window_$(model) @..\window The macros "model" and "local" are defined by the file "MAKEFILE" in each object directory. An example of the file "MAKEFILE" in the medium memory model object directory is: # # !include example # model = m # memory model required local = # memory model specific options !include ..\window.mif Notice that changes that affect all the memory models may be made in one file, namely "WINDOW.MIF". Any changes that are specific to a memory model may be made to the "MAKEFILE" in the object directory. To update the medium memory model library, the following commands may be executed: Example: C>cd \window\bcsd C>wmake A DOS ".BAT" or OS/2 ".CMD" file may be used to update all the different memory models. If the following DOS "MAKEALL.BAT" (OS/2 "MAKEALL.CMD") file is located somewhere in the "PATH", we may update all the libraries. cd \window\scsd wmake %1 %2 %3 %4 %5 %6 %7 %8 %9 cd \window\scbd wmake %1 %2 %3 %4 %5 %6 %7 %8 %9 cd \window\bcsd wmake %1 %2 %3 %4 %5 %6 %7 %8 %9 cd \window\bcbd wmake %1 %2 %3 %4 %5 %6 %7 %8 %9 cd \window\bchd wmake %1 %2 %3 %4 %5 %6 %7 %8 %9 The batch file parameters are useful if you want to specify options to Make. For instance, a global recompile may be done by executing: Example: C>makeall /a The !include preprocessing directive is a good way to partition common information so that it may be maintained easily. Another use of the !include involves program generated makefile information. For instance, if we have a program called "WMKMK" that will search through source files and generate a file called "WMKMK.MIF" that contains: # # program generated makefile information # C_to_OBJ = $(compiler) $[* $(compile_options) OBJECTS = WINDOW.OBJ BIOS.OBJ KEYBOARD.OBJ MOUSE.OBJ WINDOW.OBJ : ..\C\WINDOW.C ..\H\PROTO.H ..\H\GLOBALS.H $(C_to_OBJ) BIOS.OBJ : ..\C\BIOS.C ..\H\BIOS_DEF.H ..\H\GLOBALS.H $(C_to_OBJ) KEYBOARD.OBJ : ..\C\KEYBOARD.C ..\H\PROTO.H ..\H\GLOBALS.H $(C_to_OBJ) MOUSE.OBJ : ..\C\MOUSE.C ..\H\PROTO.H ..\H\GLOBALS.H $(C_to_OBJ) In order to use this program generated makefile information, we use a "MAKEFILE" containing: # # makefile that makes use of generated makefile information # compile_options = /mf /d1 /w3 first_target : window.lib .SYMBOLIC echo done !include wmkmk.mif window.lib : $(OBJECTS) wlib window $(OBJECTS) make : .SYMBOLIC wmkmk /r ..\c\*.c+..\c\*.cpp+..\h Notice that there is a symbolic target "first_target" that is used as a "place holder". The default behaviour for Make is to "make" the first target encountered in the makefile. The symbolic target "first_target" ensures that we have control over what file will be updated first (in this case "WINDOW.LIB"). The use of the !include preprocessing directive simplifies the use of program generated makefile information because any changes are localized to the file "MAKEFILE". As program development continues, the file "WMKMK.MIF" may be regenerated so that subsequent invocations of WMAKE benefit from the new makefile information. The file "MAKEFILE" even contains the command to regenerate the file "WMKMK.MIF". The symbolic target "make" has an associated command list that will regenerate the file "WMKMK.MIF". The command list can be executed by typing the following command: Example: C>wmake make The use of the !include preprocessing directive is a simple way to reduce maintenance of related makefiles. Hint: Macros are expanded on !include preprocessor control lines. This allows many benefits like: !include $(%env_var) so that the files that Make will process can be controlled through many different avenues like internal macros, command line macros, and environment variables. Another way to access files is through the suffix path feature of Make. A definition like .mif : c:\mymifs;d:\some\more\mifs will cause Make to search different paths for any make include files. ═══ 8.16.2. Conditional Processing ═══ Watcom Make has conditional preprocessing directives available that allow different declarations to be processed. The conditional preprocessing directives allow the makefile to 1. check whether a macro is defined, and 2. check whether a macro has a certain value. The macros that can be checked include 1. normal macros "$()" 2. environment macros "$(%)" The conditional preprocessing directives allow a makefile to adapt to different external conditions based on the values of macros or environment variables. We can define macros on the WMAKE command line as shown in the following example. Example: C>wmake "macro=some text with spaces in it" Alternatively, we can include a makefile that defines the macros if all the macros cannot fit on the command line. This is shown in the following example: Example: C>wmake /f macdef.mif /f makefile Also, environment variables can be set before WMAKE is invoked. This is shown in the following example: Example: C>set macro=some text with spaces in it C>wmake Now that we know how to convey information to Make through either macros or environment variables, we will look at how this information can be used to influence makefile processing. Make has conditional preprocessing directives that are similar to the C preprocessor directives. Make supports these preprocessor directives: !ifeq !ifneq !ifeqi !ifneqi !ifdef !ifndef along with !else !endif Together these preprocessor directives allow selection of makefile declarations to be based on either the value or the existence of a macro. Environment variables can be checked by using an environment variable name prefixed with a "%". A common use of a conditional preprocessing directive involves setting environment variables. # # setting an environment variable # !ifndef %lib .BEFORE set lib=c:\watcom\lib386;c:\watcom\lib386\dos !endif If you are writing portable applications, you might want to have: # # checking a macro # !include version.mif !ifdef OS2 machine = /2 # compile for 286 !else machine = /0 # default: 8086 !endif The !ifdef ("if defined") and !ifndef ("if not defined") conditional preprocessing directives are useful for checking boolean conditions. In other words, the !ifdef and !ifndef are useful for "yes-no" conditions. There are instances where it would be useful to check a macro against a value. In order to use the value checking preprocessor directives, we must know the exact value of a macro. A macro definition is of the form: = Make will first strip any comment off the line. The macro definition will then be the text following the equal "=" sign with leading and trailing blanks removed. Initially this might not seem like a sensible way to define a macro but it does lend itself well to defining macros that are common in makefiles. For instance, it allows definitions like: # # sample macro definitions # link_options = debug line # line number debugging compile_options = /d1 /s # line numbers, no stack checking These definitions are both readable and useful. A makefile can handle differences between compilers with the !ifeq, !ifneq, !ifeqi and !ifneqi conditional preprocessing directives. The first two perform case sensitive comparisons while the last two perform case insensitive comparisons. One way of setting up adaptive makefiles is: # # options made simple # compiler = wcc386 stack_overflow = No # yes -> check for stack overflow line_info = Yes # yes -> generate line numbers !ifeq compiler wcc386 !ifneqi stack_overflow yes stack_option = /s !endif !ifeqi line_info yes line_option = /d1 !endif !endif !ifeq compiler tcc !ifeqi stack_overflow yes stack_option = -N !endif !ifeqi line_info yes line_option = -y !endif !endif # # make sure the macros are defined # !ifndef stack_option stack_option = !endif !ifndef line_option line_option = !endif example : .SYMBOLIC echo $(compiler) $(stack_option) $(line_option) The conditional preprocessing directives can be very useful to hide differences, exploit similarities, and organize declarations for applications that use many different programs. Another directive is the !define directive. This directive is equivalent to the normal type of macro definition (i.e., macro = text) but will make C programmers feel more at home. One important distinction is that the !define preprocessor directive may be used to reflect the logical structure of macro definitions in conditional processing. For instance, the previous makefile could have been written in this style: !ifndef stack_option ! define stack_option !endif !ifndef line_option ! define line_option !endif The "!" character must be in the first column but the directive keyword can be indented. This freedom applies to all of the preprocessing directives. The !else preprocessing directive benefits from this type of style because !else can also check conditions like: !else ifeq !else ifneq !else ifeqi !else ifneqi !else ifdef !else ifndef so that logical structures like: !ifdef %version ! ifeq %version debugging ! define option = debug all ! else ifeq %version beta ! define option = debug line ! else ifeq %version production ! define option = debug ! else ! error invalid value in VERSION ! endif !endif can be used. The above example checks the environment variable "VERSION" for three possible values and acts accordingly. Another derivative from the C language preprocessor is the !error directive which has the form of !error in Make. This directive will print out the text and terminate processing of the makefile. It is very useful in preventing errors from macros that are not defined properly. Here is an example of the !error preprocessing directive. !ifndef stack_option ! error stack_option is not defined !endif !ifndef line_option ! error line_option is not defined !endif There is one more directive that can be used in a makefile. The !undef preprocessing directive will clear a macro definition. The !undef preprocessing directive has the form: !undef The macro identifier can represent a normal macro or an environment variable. A macro can be cleared after it is no longer needed. Clearing a macro will reduce the memory requirements for a makefile. If the macro identifier represents an environment variable (i.e., the identifier has a "%" prefix) then the environment variable will be deleted from the current environment. The !undef preprocessing directive is useful for deleting environment variables and reducing the amount of internal memory required during makefile processing. ═══ 8.16.3. Loading Dynamic Link Libraries ═══ Watcom Make supports loading of Dynamic Link Library (DLL) versions of Watcom software through the use of the !loaddll preprocessing directive. This support is available on Win32 and 32-bit OS/2 platforms. Performance is greatly improved by avoiding a reload of the software for each file to be processed. The syntax of the !loaddll preprocessing directive is: !loaddll $(exename) $(dllname) where $(exename) is the command name used in the makefile and $(dllname) is the name of the DLL to be loaded and executed in its place. For example, consider the following makefile which contains a list of commands and their corresponding DLL versions. # Default compilation macros for sample programs # # Compile switches that are enabled CFLAGS = -d1 CC = wpp386 $(CFLAGS) LFLAGS = DEBUG ALL LINK = wlink $(LFLAGS) !ifdef __LOADDLL__ ! loaddll wcc wccd ! loaddll wccaxp wccdaxp ! loaddll wcc386 wccd386 ! loaddll wpp wppdi86 ! loaddll wppaxp wppdaxp ! loaddll wpp386 wppd386 ! loaddll wlink wlink ! loaddll wlib wlibd !endif .c.obj: $(CC) $*.c The __LOADDLL__ symbol is defined for versions of Watcom Make that support the !loaddll preprocessing directive. The !ifdef __LOADDLL__ construct ensures that the makefile can be processed by an older version of Watcom Make. Make will look up the wpp386 command in its DLL load table and find a match. It will then attempt to load the corresponding DLL (i.e., wppd386.dll) and pass it the command line for processing. The lookup is case insensitive but must match in all other respects. For example, if a path is included with the command name then the same path must be specified in the !loaddll preprocessing directive. This problem can be avoided through the use of macros as illustrated below. # Default compilation macros for sample programs # # Compile switches that are enabled # cc286 = wpp cc286d = wppdi86 cc386 = wpp386 cc386d = wppd386 linker = wlink linkerd = wlink CFLAGS = -d1 CC = $(cc386) $(CFLAGS) LFLAGS = DEBUG ALL LINK = wlink $(LFLAGS) !ifdef __LOADDLL__ !loaddll $(cc286) $(cc286d) !loaddll $(cc386) $(cc386d) !loaddll $(linker) $(linkerd) !endif .c.obj: $(CC) $*.c A path and/or extension may be specified with the DLL name if desired. ═══ 8.17. Command List Directives ═══ Watcom Make supports special directives that provide command lists for different purposes. If a command list cannot be found while updating a target then the directive .DEFAULT may be used to provide one. A simple .DEFAULT command list which makes the target appear to be updated is: .DEFAULT wtouch $^@ The Watcom Touch utility sets the time-stamp on the file to the current time. The effect of the above rule will be to "update" the file without altering its contents. In some applications it is necessary to execute some commands before any other commands are executed and likewise it is useful to be able to execute some commands after all other commands are executed. Make supports this capability by checking to see if the .BEFORE and .AFTER directives have been used. If the .BEFORE directive has been used, the .BEFORE command list is executed before any commands are executed. Similarly the .AFTER command list is executed after processing is finished. It is important to note that if all the files are up to date and no commands must be executed, the .BEFORE and .AFTER command lists are never executed. If some commands are executed to update targets and errors are detected (non-zero return status, macro expansion errors), the .AFTER command list is not executed (the .ERROR directive supplies a command list for error conditions and is discussed in this section). These two directives may be used for maintenance as illustrated in the following example: # # .BEFORE and .AFTER example # .BEFORE echo .BEFORE command list executed .AFTER echo .AFTER command list executed # # rest of makefile follows # . . . If all the targets in the makefile are up to date then neither the .BEFORE nor the .AFTER command lists will be executed. If any of the targets are not up to date then before any commands to update the target are executed, the .BEFORE command list will be executed. The .AFTER command list will be executed only if there were no errors detected during the updating of the targets. The .BEFORE, .DEFAULT, and .AFTER command list directives provide the capability to execute commands before, during, and after the makefile processing. Make also supports the .ERROR directive. The .ERROR directive supplies a command list to be executed if an error occurs during the updating of a target. # # .ERROR example # .ERROR beep # # rest of makefile follows # . . . The above makefile will audibly signal you that an error has occurred during the makefile processing. If any errors occur during the .ERROR command list execution, makefile processing is terminated. ═══ 8.18. MAKEINIT File ═══ As you become proficient at using Watcom Make, you will probably want to isolate common makefile declarations so that there is less duplication among different makefiles. Make will search for a file called "MAKEINIT" and process it before any other makefiles. The search for the "MAKEINIT" file will occur along the current "PATH". If the file "MAKEINIT" is not found, processing continues without any errors. The only default declaration that Make provides is equivalent to a "MAKEINIT" file containing: __MAKEOPTS__ = __MAKEFILES__ = __MSDOS__ = # clear .EXTENSIONS list .EXTENSIONS: # set .EXTENSIONS list .EXTENSIONS: .exe .exp .lib .obj .asm .c .for .pas .cob .h .fi .mif For OS/2, the __MSDOS__ macro will be replaced by __OS2__ and for Windows NT, the __MSDOS__ macro will be replaced by __NT__. The use of a "MAKEINIT" file will allow you to reuse common declarations and will result in simpler, more maintainable makefiles. ═══ 8.19. Command List Execution ═══ Watcom Make is a program which must execute other programs and operating system shell commands. There are three basic types of executable files in DOS. 1. .COM files 2. .EXE files 3. .BAT files There are two basic types of executable files in Windows NT. 1. .EXE files 2. .BAT files There are two basic types of executable files in OS/2. 1. .EXE files 2. .CMD files The .COM and .EXE files may be loaded into memory and executed. The .BAT files must be executed by the DOS command processor or shell, "COMMAND.COM". The .CMD files must be executed by the OS/2 command processor or shell, "CMD.EXE" Make will search along the "PATH" for the command and depending on the file extension the file will be executed in the proper manner. If Make detects any input or output redirection characters (these are ">", "<", and "|") in the command, it will be executed by the shell. Under DOS, an asterisk prefix (*) will cause Make to examine the length of the command argument. If it is too long (> 126 characters), it will take the command argument and stuff it into a temporary environment variable and then execute the command with "@env_var" as its argument. Suppose the following sample makefile fragment contained a very long command line argument. # # Asterisk example # *foo myfile /a /b /c ... /x /y /z Make will perform something logically similar to the following steps. set TEMPVAR001=myfile /a /b /c ... /x /y /z foo @TEMPVAR001 The command must, of course, support the "@env_var" syntax. Typically, DOS commands do not support this syntax but many of the Watcom tools do. The exclamation mark prefix (!) will force a command to be executed by the shell. Also, the command will be executed by the shell if the command is an internal shell command from the following list: break (check for Ctrl+Break) call (nest batch files) chdir (change current directory) cd (change current directory) cls (clear the screen) cmd (start NT or OS/2 command processor) command (start DOS command processor) copy (copy or combine files) ctty (DOS redirect input/output to COM port) d: (change drive where "d" represents a drive specifier) date (set system date) del (erase files) dir (display contents in a directory) echo (display commands as they are processed) erase (erase files) for (repetitively process commands, intercepted by WMAKE) if (allow conditional processing of commands) md (make directory) mkdir (make directory) path (set search path) pause (suspend batch operations) prompt (change command prompt) ren (rename files) rename (rename files) rmdir (remove directory) rd (remove directory) set (set environment variables, intercepted by WMAKE) time (set system time) type (display contents of a file) ver (display the operating system version number) verify (set data verification) vol (display disk volume label) The operating system shell "SET" command is intercepted by Make. The "SET" command may be used to set environment variables to values required during makefile processing. The environment variable changes are only valid during makefile processing and do not affect the values that were in effect before Make was invoked. The "SET" command may be used to initialize environment variables necessary for the makefile commands to execute properly. The setting of environment variables in makefiles reduces the number of "SET" commands required in the system initialization file. Here is an example with the Watcom C/C++ compiler. # # SET example # .BEFORE set include=c:\special\h;$(%include) set lib=c:\watcom\lib386;c:\watcom\lib386\dos # # rest of makefile follows # . . . The first "SET" command will set up the INCLUDE environment variable so that the Watcom C/C++ compiler may find header files. Notice that the old value of the INCLUDE environment variable is used in setting the new value. The second "SET" command indicates to the Watcom Linker that libraries may be found in the indicated directories. Environment variables may be used also as dynamic variables that may communicate information between different parts of the makefile. An example of communication within a makefile is illustrated in the following example. # # internal makefile communication # .BEFORE set message=message text 1 echo *$(%message)* set message= echo *$(%message)* .example : another_target .SYMBOLIC echo *$(%message)* another_target : .SYMBOLIC set message=message text 2 The output of the previous makefile would be: (command output only) *message text 1* ** *message text 2* Make handles the "SET" command so that it appears to work in an intuitive manner similar to the operating system shell's "SET" command. The "SET" command also may be used to allow commands to relay information to commands that are executed afterwards. The DOS "FOR" command is intercepted by Make. The reason for this is that DOS has a fixed limit for the size of a command thus making it unusable for large makefile applications. One such application that can be done easily with Make is the construction of a WLINK command file from a makefile. The idea behind the next example is to have one file that contains the list of object files. Anytime this file is changed, say, after a new module has been added, a new linker command file will be generated which in turn, will cause the linker to relink the executable. First we need the makefile to define the list of object files, this file is "OBJDEF.MIF" and it declares a macro "objs" which has as its value the list of object files in the application. The content of the "OBJDEF.MIF" file is: # # list of object files # objs = & window.obj & bios.obj & keyboard.obj & mouse.obj The main makefile ("MAKEFILE") is: # # FOR command example # !include objdef.mif plot.exe : $(objs) plot.lnk wlink @plot plot.lnk : objdef.mif echo NAME $^& >$^@ echo DEBUG all >>$^@ for %i in ($(objs)) do echo FILE %i >>$^@ This makefile would produce a file "PLOT.LNK" automatically whenever the list of object files is changed (anytime "OBJDEF.MIF" is changed). For the above example, the file "PLOT.LNK" would contain: NAME plot DEBUG all FILE window.obj FILE bios.obj FILE keyboard.obj FILE mouse.obj Make supports eight internal commands: 1. %null 2. %stop 3. %quit 4. %abort 5. %create 6. %write 7. %append 8. %make The %null internal command does absolutely nothing. It is useful because Make demands that a command list be present whenever a target is updated. # # %null example # all : application1 application2 .SYMBOLIC %null application1 : appl1.exe .SYMBOLIC %null application2 : appl2.exe .SYMBOLIC %null appl1.exe : (dependents ...) (commands) appl2.exe : (dependents ...) (commands) Through the use of the %null internal command, multiple application makefiles may be produced that are quite readable and maintainable. The %stop internal command will temporarily suspend makefile processing and print out a message asking whether the Makefile processing should continue. Make will wait for either the "y" key (indicating that the Makefile processing should continue) or the "n" key. If the "n" key is pressed, makefile processing will stop. The %stop internal command is very useful for debugging makefiles but it may be used also to develop interactive makefiles. The %quit internal command will terminate execution of Make and return to the operating system shell with an exit code of zero. The %abort internal command is identical to %quit except that a non-zero exit code is returned by WMAKE. The %create, %write, and %append internal commands allow WMAKE to generate files under makefile control. This is useful for files that have contents that depend on makefile contents. Through the use of macros and the "for" command, Make becomes a very powerful tool in maintaining lists of files for other programs. The %create internal command will create or truncate a file so that the file does not contain any text. The %create internal command has the form: %create where is a file specification. The %write internal command will create or truncate a file and write one line of text into it. The %append internal command will append a text line to the end of a file (which will be created if it does not exist). The %write and %append internal commands have the same form, namely: %write %append where is a file specification and is arbitrary text. Full macro processing is performed on these internal commands so the full power of WMAKE can be used. The following example illustrates a common use of these internal commands. # # %create %append example # !include objdef.mif plot.exe : $(objs) plot.lnk wlink @plot plot.lnk : objdef.mif %create $^@ %append $^@ NAME $^& %append $^@ DEBUG all for %i in ($(objs)) do %append $^@ FILE %i The above code demonstrates a valuable technique that can generate directive files for WLINK, WLIB, and other utilities. The %make internal command permits the updating of a specific target. The %make internal command has the form: %make where is a target in the makefile. # # %make example # !include objdef.mif plot.exe : $(objs) %make plot.lnk wlink @plot plot.lnk : objdef.mif %create $^@ %append $^@ NAME $^& %append $^@ DEBUG all for %i in ($(objs)) do %append $^@ FILE %i ═══ 8.20. Compatibility Between Watcom Make and UNIX Make ═══ Watcom Make was originally based on the UNIX Make utility. The PC's operating environment presents a base of users which may or may not be familiar with the UNIX operating system. Make is designed to be a PC product with some UNIX compatibility. The line continuation in UNIX Make is a backslash ("\") at the end of the line. The backslash ("\") is used by the operating system for directory specifications and as such will be confused with line continuation. For example, you could type: cd \ along with other commands ... and get unexpected results. However, if your makefile does not contain path separator characters ("\") and you wish to use "\" as a line continuation indicator then you can use the Make "u" (UNIX compatibility mode) option. Also, in the UNIX operating system there is no concept of file extensions, only the concept of a file suffix. Make will accept the UNIX Make directive .SUFFIXES for compatibility with UNIX makefiles. The UNIX compatible special macros supported are: $@ full name of the target $* target with the extension removed $< list of all dependents $? list of dependents that are younger than the target The extra checking of makefiles done by Make will require modifications to UNIX makefiles. The UNIX Make utility does not check for the existence of targets after the associated command list is executed so the "c" or the .NOCHECK directive should be used to disable this checking. The lack of a command list to update a target is ignored by the UNIX Make utility but Watcom Make requires the special internal command %null to specify a null command list. In summary, Make supports many of the features of the UNIX Make utility but is not 100% compatible. ═══ 8.21. Watcom Make Diagnostic Messages ═══ This section lists the various warning and error messages that may be issued by the Watcom Make. In the messages below, %? character sequences indicate places in the message that are replaced with some other string. 1 Out of memory 2 Make execution terminated 3 Option %c%c invalid 4 %c%c must be followed by a filename 5 No targets specified 6 Ignoring first target in MAKEINIT 7 Expecting a %M 8 Invalid macro name %E 9 Ignoring out of place %M 10 Macros nested too deep 11 Unknown internal command 12 Program name is too long 13 No control characters allowed in options 14 Cannot execute %E: %Z 15 Syntax error in %s command 16 Nested %s loops not allowed 17 Token too long, maximum size is %d chars 18 Unrecognized or out of place character '%C' 19 Target %E already declared %M 20 Command list does not belong to any target 21 Extension(s) %E not defined 22 No existing file matches %E 23 Extensions reversed in implicit rule 24 More than one command list found for %E 25 Extension %E declared more than once 26 Unknown preprocessor directive: %s 27 Macro %E is undefined 28 !If statements nested too deep 29 !%s has no matching !if 30 Skipping !%1 block after !%2 31 %1 not allowed after !%2 32 Opening file %E: %Z 34 !%s pending at end of file 35 Trying to !%s an undefined macro 36 Illegal attempt to update special target %E 37 Target %E is defined recursively 38 %E does not exist and cannot be made from existing files 39 Target %E not mentioned in any makefile 40 Could not touch %E 41 No %s commands for making %E 42 Last command making (%L) returned a bad status 43 Deleting %E: %Z 44 %s command returned a bad status 45 Maximum string length exceeded 46 Illegal character value %xH in file 47 Assuming target(s) are .%s 48 Maximum %%make depth exceeded 49 Opening (%s) for write: %Z 50 Unable to write: %Z 51 CD'ing to %E: %Z 52 Changing to drive %C: 53 DOS memory inconsistency detected! System may halt ... 53 OS corruption detected 54 While reading (%s): %Z 59 !IF Parse Error 60 TMP Path /File Too Long 61 Unexpected End of File 62 Only NO(KEEP) allowed here 63 Non-matching " 64 Invalid String Macro Substitution 65 File Name Length Exceeded 66 Redefinition of .DEFAULT Command List 67 Non-matching { In Implicit Rule 68 Invalid Implicit Rule Definition 69 Path Too Long 70 Cannot Load/Unload DLL %E 71 Initialization of DLL %E returned a bad status 72 DLL %E returned a bad status 73 Illegal Character %C in macro name 74 in closing file %E 75 in opening file %E 76 in writing file %E 77 User Break Encountered 78 Error in Memory Tracking Encountered 79 Makefile may be Microsoft try /ms switch ═══ 9. The Touch Utility ═══ This chapter describes the Watcom Touch utility. Watcom Touch will set the time-stamp (i.e., the modification date and time) of one or more files. The new modification date and time may be the current date and time, the modification date and time of another file, or a date and time specified on the command line. This utility is normally used in conjunction with the Watcom Make utility. The rationale for bringing a file up-to-date without altering its contents is best understood by reading the chapter which describes the Make utility. The Watcom Touch command line syntax is: WTOUCH [options] file_spec [file_spec...] The square brackets [ ] denote items which are optional. options is a list of valid options, each preceded by a slash ("/") or a dash ("-"). Options may be specified in any order. file_spec is the file specification for the file to be touched. Any number of file specifications may be listed. The wild card characters "*" and "?" may be used. The following is a description of the options available. c do not create an empty file if the specified file does not exist d specify the date for the file time-stamp in "mm-dd-yy" format f use the time-stamp from the specified file i increment time-stamp before touching the file q suppress informational messages r touch file even if it is marked read-only t