Night Owl 25

home *** CD-ROM | disk | FTP | other *** search

/ Night Owl 25 / nopv25.iso / 040A / TMTP100D.ZIP / MANUAL.DOS < prev next >

Wrap

Text File | 1997-04-02 | 41.3 KB | 1,320 lines

Pascal Lite Compiler for MS DOS [ Version 1.00 ] TMT Development Corporation [[[ SEARCH THIS FILE FOR 0.nn FOR THE NEW VERSION INFORMATION ]]] Introduction ------------ The TMT Pascal Lite compiler is a fast compiler for the Pascal language. The compiler emits 32-bit code and supports many language extensions from Borland Pascal (BP), as well as more powerful extensions. The compiler requires at least 2Mb of free extended memory (XMS) and a floating point co-processor. This manual consists of the following parts: - introduction - free version limitations - compiler installation and running - configuration file - memory organization - calling convention - implemented language description: - restriction - extensions - incompatibilities - units description: - DOS - CRT - USE32 - ERRCODES - DEBUG - STRINGS - version 0.33 changes - possible problems during usage - run_time error codes Free Version Limitations---Getting an unrestricted Version ---------------------------------------------------------- The free version of the PLT compiler limits the size of the executible program code+data+heap to approximately 1.5mb. If this restriction prevents you from implementing your application, you should obtain a licensed copy of the compiler. The other advantages of ordering an unrestricted copy of the compiler are: 1> IDE 2> Debugger 3> Full on-line documentation. The special price of the unrestricted version of the compiler is only US$75. The distribution is electronic; if you desire your software to be shipped by regular mail, please contact sales@tmt.com for details. Unrestricted license will get you the current version of the compiler plus all updates for up to a year as they are released. To order your copy of PLT/Lite, send an email to sales@tmt.com with the credit card information (VISA, MC, AMEX, DISCOVER). Alternatively, call 1-718-268-3756 (10-5 EST) or send check or postal order to TMT Development Corp. 19 POPLAR AVE, Stony Brook, NY 11790-1751 Foreign orders: if you are paying by check, please make sure that your check is in US dollars and drawn on a US bank. All postal orders: please make sure that your order specifies a contact email account. PO's: we will accept PO's from qualified institutions in US and Canada. Please verify that your PO will be accepted before issuing it. Compiler installation and running --------------------------------- The Pascal Lite distribution is contained in the tmtpldos.zip archive. Please, read the license.doc file before the installation. Follow these instructions to run the installation process. - create the \TMTPL directory on one of the drives (we will call it the x: drive) using the command: x> md TMTPL - copy the tmtp100d.zip file into the TMTPL directory: x> copy tmtp100d.zip x:\TMTPL x> x: x> cd \TMTPL - run the PKUNZIP program to decompress the tmtp100d.zip x:\TMTPL> pkunzip -d tmtp100d (the switch "-d" unpacks with subdirectories) - modify the PATH statement in the autoexec.bat, to include the x:\TMTPL\BIN subdirectory, if you so desire. - goto to \TMTPL\EXAMPLES\DOS\HELLO compile the simple test program, "HELLO.PAS", and run it: x:\TMTPL> plt hello x:\TMTPL> hello - read the entire manual.doc file before using the compiler for any "real" application development. Contents of the distribution ---------------------------- documentation: manual.doc - description of the PLT compiler license.doc - license agreement readme - this file executable: subdirectory BIN contains following files: plt.cfg - compiler configuration file dos32.exe - the DOS32 protected mode extender dos4gw.exe - the DOS/4GW protected mode extender (used by compiler) plt.exe - the PLT compiler under the DOS/4GW extender pltpmw.exe - the PLT compiler under the PMODE extender standard library object modules: subdirectory UNITS\ contains unit obj modules files with .FPD extensions arg - commmand line parsing unit crt - crt unit debug - post-mortem dump unit and other debugging functions dos.fpd - dos unit doscall - low-level DOS interface errcodes - run-time error codes strings - strings unit system - system unit also in this directory contains files: DOS32.EXE - DOS32 extender STUB32.EXE - stub for DOS32 extender PASSTUB.EXE - PMODE-based extender example: subdirectory EXAMPLE\ contains following examples for MS DOS: HELLO\ - example "Hello, World" program FLAME\ - example of direct works with I/O ports and physical memory VESADEMO\ - example of import and usage VESA.OBJ file from DOS32 libraries. +++0.33 LIN_EQ\ - example of multidimensional open arrays usage ---0.33 supplemential files: DOS32\ - subdirectory that contains the DOS32 distribution to be used by the application. Running the compiler: -------------------- The compiler is invoked with the "plt" command: plt [switches] <main file> where the following switches can be used: -m - make modified units (default option) -b - build the whole program -c - single compilation (without linkage) After that the suite of object files with .FPD extensions will be established in the directory with the sources, and the executable file will be created in the current directory (if the switch -c is not set). Instead of the "plt.exe" file there is another version of the compiler which uses the freeware extender PMODE. This is a "pltpmw.exe" file. This compiler demands less memory for the compilation and works faster then "plt.exe". Try to use it. If problems arise, use "plt.exe". Out of memory message If you receive message "not enough memory for compilation", you have to either: - increase a amount of free extended memory (for example, decrease the "cash" of your hard disk), or - decrease the value of the EXEMAK and OBJMAX parameters in the configuration file, or - switch on the virtual memory of DOS/4GW: To do this, include a line "SET DOS4GW=1" into the your AUTOEXEC.BAT and use the compiler version under the DOS/4GW extender. Configuration file ------------------ Before a compilation the compiler reads the settings from the special file - plt.cfg. It searches for one at the beginning of the current directory and then, if it doesn't find one, in the starting directory. The configuration file "plt.cfg" contains the compilation parameters: one command per line. Empty line are possible. Comment line are started with the ';' symbol. The following parameters can be used in the plt.cfg: toggle settings: w+ - turn on the puting of warning messages r+ - turn on the range checking q+ - turn on the overflow checking i+ - turn on then automatic input/output result checking t+ - typed pointers x+ - extended syntax v+ - strict var checking optreg+ - turn on the register optimization optfrm+ - turn on the stack frame optimization opt+ - turn on all optimization (optreg+ & optfrm+) Default settings: w+, r+, q+, i+, t-, v+, x+, opt+. file path search specification: srcpath <search_path> - specifies path for the source file search objpath <search_path> - specifies path for the object file (.FPD) and STUBs search objimppath <search_path> - specifies path for the OBJ import file search by use of a $L directive pseudo-devices SYS: and SRC: can be used in the path line: SYS: means the PLT compiler directory, SRC: means the application program directory. defaults: srcpath src:;sys: objpath src:;sys: objimppath src:;sys: buffer size specification: objmax <size> - specifiies the size of the buffer for object modules (fpd-files). This parameter must be about one and a half times the size of the largest fpd-file from the project. exemax <size> - specifies the maximum size of the executable module. defaults: objmax 333000 exemax 256000 These values are enough for ordinary work. But if you receive a message about overflowing of these buffers, you'll have to increase them. If you receive the message about unsufficient memory for the compilation, you can try to decrease them. stack size specification: stack <stack_size> - specifies size of the application stack default: 32000 start up block (stub) name specification: stub <stub_name> - specifies the STUB name for the linker. default: DOS32.EXE +++0.30 use the OBJPATH option to specify the seacrh path for the stub files ---0.30 Symbol definition: def <symbol name> +++0.30 The following symbols are predefined: __TMT__ : always __DOS__ : for DOS target * __OS2__ : for all OS/2 targets * __DLL__ : for .DLL targets * __PM__ : for OS/2 Presentation manager targets * __FS__ : for OS/2 Full screen targets * -- OS/2-hosted compiler only ---0.30 error message format specification: errmw+ - specifies the MetaWare Pascal format of the error message: E "HELLO.PAS" L4/C16: type mismatch errmw- - specifies BP-like format of the error message: textcolor (lightred+'a'); Error: HELLO.PAS (line 4, col 16): type mismatch This parameter allows use of the MultiEdit program as a shell when the errmw+ parameter is set. The default is: errmw- Output of the Dos32 logo specification: logo+ - enables output of the Dos32 logo during the running of the emitted program. logo- - disables output of the Dos32 logo during the running of the emitted program. The default is: logo+ Memory organization ------------------- The PLT compiler uses the Dos32 extender for a protected-mode program. The full distributive of Dos32 is in the DOS32 subdirectory. This distributive contains all information about program execution under the 32-bit mode. Here some important issues are presented. The segment registers are not used in the protected mode. Instead all address space is separated on 4Kb pages starting from the zero address. Every page has a physical equivalent in the special table (the physical address might not be equal to the logical address). In such a case the logical address 0 corresponds to the beginning of the program code. Therefore if you try to access the physical address (video memory and so on), you will not succeed. Nevertheless, access to the physical addresses is possible, because the Dos32 extender reflects physical addresses into another set of special address pages. To find the logical address from physical, you have to add the value of the special _zero varible from the SYSTEM unit to the physical address. For example: procedure clr_video (filler: char); var i: integer; begin for i := 0 to 80*25-1 do mem [_zero+$B8000+i*2] := filler; end; This procedure fills the video memory of the VGA adapter with the filler symbol. Note that the linear address $B8000 is used as the physical address - not the segment address $B800. Some another special variables are described in the SYSTEM unit. The _psp variable contains the logical 32-bit address of the PSP of the program, and the _environ variable contains the environment address. Although you can access the interrupt vectors by using this method, we do not suggest doing this. Also keep in mind that MS-DOS interrupt handlers use memory addresses in the 1st mb of physical memory while your program and its data are loaded beyond the 1st mb. DOS32 intercepts and correctly handles some, but not all, calls to MS-DOS. Thus, if you are using Intr() or MsDos() calls, or call MS-DOS from the assembler, you will need to modify the code. Additional details about this and related subjects can be found in DOS32.DOC and API.DOC in subdirectory DOS32. Calling conventions ------------------- Calling conventions match those in Borland Pascal with the following differences: all parameters use 4 bytes on the stack, or a multiple of 4 (BP:2) all procedures must preserve the contents of registers ebx, ecx, edx, ds, and es! the direction bit should be cleared after the exit from a procedure, if it has been modified by it. Language syntax of TMT Pascal/Fast: ----------------------------------- While the Language syntax of TMT Pascal/Fast is in general compatible with BP 7.0, there are some differences. Below is the list, divided into three groups: limitations, extensions, and incompatibilities. 1. Limitations: Not implemented are MARK and RELEASE. The INLINE() operator in implemented in a partial form: INLINE (byte/byte/...); (No references to variables/constants are allowed). Import of object modules does not support all 32bit object formats. We recommend using TASM which is fully supported, excepting usage of SEG addresses. 2. Extensions: ADA-style comments are supported: -- comment For example: space := ' '; -- initialize filler char Almost everywhere where Pascal's syntax allows a type identifier, we also allow a type descriptor. The compiler will produce a warning if you use this feature. !!! Keep in mind that the rules of type equivalence stay in force. Procedural values contain the address of the local environment (frame). Thus, any local procedures can be used in procedural values. Procedural declaration in the style procedure p (function f (a:real):real); ... are also realized. The procedural value from a method of object can be obtained by selecting this method from some object value (not from a type). The parameters of this procedural value must match the parameters of the method. The invocation of such a procedural value is an invocation of the corresponding method of the object. The reference to the object is transferred through the base of the procedural value. You can use only global procedural values to initialize a type constant. !!! Procedural values may be used only while the environment where they were formed is still in existence. Thus, for local procedures -- until the exit from the block, in which they are described; for methods -- while the underlying object still exists. see also "incompatibilities." With TMT Pascal you can use any statement as a procedure body, except for the assignment and the procedure calls. The RESULT variable in the body of such functions denotes the variable that contains the return value. The RESULT is of the function return type and may be used as a variable without any restrictions. With TMT Pascal you can enter the procedure body directly as a procedure parameter. The procedure or function header (if not specified) takes the procedural parameter type. If the procedure header is specified, the procedure name is omitted. Example: function integral (function f (a:real):real; low, high, step: real): real; begin ... end; ... writeln (integral ( function (x:real):real; begin result := sqrt (x) end, 0, 10, 0.1)); writeln (integral (begin result := sqrt (a) end, 0, 10, 0.1)); writeln (integral ( function; -- function keyword needed var x: real; -- for local declaration begin x := sqrt (a); result := x end, 0, 10, 0.1)); writeln (integral ( declare; -- other way var x: real; -- for local variable declaration begin x := sqrt (a); result := x end, 0, 10, 0.1)); TMT Pascal allows exit from a local procedure to the one that contains it. This feature is listed in the Pascal's ANSI standard but not realized in BP. Together with procedural values, this is very useful for error handling: Program test; Var on_eof: procedure; Function read_char: char; Var c: char; Begin If Eof (Input) Then on_eof; read (c); read_char := c; End; Procedure p; Label eof_reached; Procedure go_eof; Begin goto eof_reached; End; Begin on_eof := go_eof; While True Do Write (read_char); Eof_reached: Writeln ('*** EOF ***'); on_eof := NIL; End; Begin p; End. Restriction: BREAK and CONTINUE operators cannot be used for exit from a procedure. Use GOTO instead. Example: for i := 1 to 10 do writeln (integral ( -- function integral from previous example if a < 0 then break else result := sqrt (a), -- incorrect i, i + 1, 0.01)); declare label L; begin for i := 1 to 10 do writeln (integral ( if a < 0 then goto L else result := sqrt (a), -- correct i, i + 1, 0.01)); L: end; Functions may return any values of any type, including structures and arrays. The new operator DECLARE DECLARE <declaration part> BEGIN <statement list> END allows one to describe local variables and procedures anywhere in the program. Use of "DECLARE" leads to more readable programs and conserves the stack space. Functions length, chr, ord may be used as variables. For example: chr (i) := ' '; is equivalent to i := ord (' '); ord (c) := 255; is equivalent to c := chr (255); length (s) := 10; is equivalent to s [0] := chr (10); The main program may contain "interface" and "implementation". This allows access to the variables of the main program from other modules: Program test; Interface Var global: Integer; Implementation Uses Unit_test; Begin Unit_test.write; End. Unit Unit_test; Interface Procedure Write_global; Implementation Uses test; Procedure Write_global; Begin Write (test.global); End; End. !!! Here the name of the file that contains the text of the main program must be identical with the name that follows the keyword "program". New type: DWORD - an unsigned 32-bit integer (same as unsigned long in C). It can be useful in many cases, but keep in mind that during arithmetic operations between DWORD's and LONGINT's both arguments are casted to LONGINT's. This may cause an error. ABSOLUTE may refer to fields of records and objects. Also, the address of a global record/object field can be used within the initialization of typed constants. Furthermore one can use recursive initialization: type rec = record next: ^rec; buffer: array [1..10] of char; buf_adr: pointer; end; const cyclic: rec = (next: @cyclic; buf_adr: @cyclic.buffer); In initialization of structures and objects the fields may go in an arbitrary order. If a field is not listed, it is zeroed up. The syntax of the STR operator has been extended: STR (value[:width[:precision]], ..., dest); Here STR (name:2, ':', ext:3, dst); is equivalent to STR (name:2, temp1); STR (':', temp2); STR (ext:3, temp3); dst := temp1 + temp2 + temp3; Negative field widths in WRITE, WRITELN, and STR left align the value within its field. !!! The Strings unit contains additional functions which may be more convenient than the STR operator. +++0.30 Multidimensional open arrays: You can now use in procedure parameters and functions descriptions "array (<dim>) of <type>", where <dim> is a positive integer constant, defining the number of dimensions, ánd <type> is the type of the array elements. To determine the upper bounds of the array, use the "high (array)" function. It returns a vector of longints ("array [0..<dim>-1] of longint") containing the upper bounds. The lower bounds are always set to 0. The vector of the lower bounds can be obtained with a similar function "low". Example: procedure print_vector (v: array (1) of double); var i: integer; begin for i := 0 to high (v) [0] do write (v [i]:10:6, ' '); writeln; end; procedure print_matrix (m: array (2) of double); var i: integer; begin for i := 0 to high (m) [0] do print_vector (m [i]); writeln; end; const a: array [1..3, 1..3] of double = ((1,0,2),(2,1,0),(1,2,1)); begin print_matrix (a); end. You can find an example program for solving a system of linear equations in the directory "EXAMPLES\LIN_EQ". ---0.30 +++0.40 NEW!!!: User defined operations. TMT Pascal allows redefining of the standard operations on predefined types and ¿ overloading of these operationg for new types. For this, we use the construction OVERLOAD: The syntax is: OVERLOAD op_sign = qualified procedure identifier; Where the op_sign is one of the standard operation symbols: + - / * = <> < > <= >= AND OR XOR SHL SHR MOD DIV IN NOT +:= -:= *:= /:= When a re-defined operation is used, TMT Pascal uses the last definition that could be applied toward operants of given types. For example, this fragment FUNCTION add2_rr (a, b: REAL): REAL; result := (a + b) * 2; FUNCTION add2_ii (a, b: INTEGER): INTEGER; result := (a + b) * 2; OVERLOAD + = add_rr; OVERLOAD + = add_ii; redefines the "+" operation. Notice that the order of OVERLOAD's is important. The reverse order OVERLOAD + = add_ii; OVERLOAD + = add_rr; will cause "add_rr" used *always* since integers can always be cast into reals. In the SOURCES subdirectory you can find the source of the COMP module which realizes the complex numbers and defines the operations on them. Operations +:=, -:=, *:= and /:= have the lowest precedence (lower, than the comparison operations) and are right-associative. The operations "+:=" and "-:=" are predefined for all integer and real types; the operations "*:=" and "/:=" are predefined for all real types, with the obvious meaning. See also description of Strings unit. NEW: Iterated typed array constats: you can use the notation "..." in the array initialization. The ellipses should be the last element in the initialization list. If used, the remaining elements of the array are filled up with the last given value that preceded the ellipses. Ellipses cannot be the first element of the initialization list. Example: const a: array [1..10, 1..10] of integer = ( (1,...), (2,...), (3,...), (4, 5, 6, ...), (0, ...), ... ); fills up the 10 by 10 array as follows: 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 4 5 6 6 6 6 6 6 6 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ---0.40 3. Incompatibilities: Procedural values is 8-byte long and has the following format: 0 +-----------------------+ | The entry point | 4 +-----------------------+ | The local environment | 8 +-----------------------+ However, a pointer to a procedure is only 4-byte long and contain only the address of the entry point. The @ operator verifies that the procedure is global. Otherwise a range error is generated. Furthermore, the stack frame structure and parameter passing conventions differ from those in BP. Thus the approach used in TVision and CLassLib for writing iterators cannot be used. However, we offer this correct and reliable (and more standard) way: Type list = object next: ^list; procedure for_all (procedure body (var v)); end; Procedure list.for_all; Var p: ^list; Begin p := @self; repeat body (p); p := p^.next; end; End; ... Type int_list = object (list) value: integer; function first_positive: ^int_list; end; Function int_list.first_positive; Label OK; Var res: ^int_list; Procedure do_item (var v); Begin If int_list (v).value > 0 Then Begin res := @v; GoTo OK; End End; Begin res := nil; for_all (do_item); OK: first_positive := res; End; ... Since in the flat model there are no segments, SEG() always returns 0, while PTR() ignores its first parameter. !!! Be careful with these functions: in practice, any use of PTR() or SEG() in your program may require a program modification. Pseudo-array Mem and MemW are used as follows: i := MemW [<linear address>] Note that the segment is not given. Furthermore, the pseudo-arrays MemD and PortD of type DWORD are available. See "Memory organization" for more details. Built-in assembler ------------------ The built-in assembler of TMT Pascal Lite is 32-bit. It is compatible with the BP's built-in assembler with the following differences: Since the program created by the Pascal Lite is executed in the flat model, the far call and jump commands as well as the @Code and @Data symbols are not implemented. The ret command: The ret command without arguments is considered as a ret <parameter block size> command. If you need to write the ret command without size you should set 0 explicitly: ret 0 Code-procedure Besides the assembler-routine you can use the code-routine. It has the following differences: the compiler doesn't emit the frame command on enter and return from the routine (including the ret command), and the local parameters are based on ESP on the moment of entry. Example: function hi (n: word); code; asm mov al, byte ptr [n+1] ret end; Supplied units: --------------- The following units are included: SYSTEM.FPD - standard library functions DOS.FPD - MS-DOS interface. Corresponds to the DOS unit of BP. CRT.FPD - Text mode screen interface. Corresponds to the CRT unit of BP. USE32.FPD - Define type Integer as 32-bit etc. ERRCODES.FPD - Run-time error codes. DEBUG.FPD - Some debugging code. STRINGS.FPD - String handling functions. Corresponds to the STRINGS unit of BP. Besides these modules, the distribution includes DOSCALL.FPD and ARG.FPD, which are needed for linkage. These modules should not be called by the user. The DOS Unit ------------ This corresponds to the DOS unit of BP. Differences: Type FarPointer = record ofs: pointer; seg: word end; The function GetIntVec has the following syntax: procedure GetIntVec (IntNo: Byte; var Vector: FarPointer); and not procedure GetIntVec (IntNo: Byte; var Vector: Pointer); New procedures: procedure GetIntVecFar (IntNo: Byte; var Vector: FarPointer); procedure SetIntVecFar (IntNo: Byte; const Vector: FarPointer); GetIntVec is equivalent to GetIntVecFar(). SetIntVec is equivalent to SetIntVecFar(), but uses in the segmented address the current CS. A header of an interrupt procedure should have the following form: procedure handler ( eip, eax, ecx, edx, ebx, esp, ebp, esi, edi: dword; gs, fs, es: word ); interrupt; The Registers type has this structure: type Registers = record edi, esi, ebp, _res : dword; case boolean of true: (ebx, edx, ecx, eax: dword; flags, es, ds, fs, gs, ip, cs, sp, ss: word); false: (bl, bh, b1, b2, dl, dh, d1, d2, cl, ch, c1, c2, al, ah: byte); end; !!! When using Intr() and MsDos(), keep in mind that the DOS interrupt handlers can deal only with the addresses from the 1st megabyte of memory. The CRT unit ------------ Corresponds to the CRT unit of BP. The USE32 unit ------------ Contains redefinitions of integer types for 32-bit computing as follow: type SmallInt = System.Integer; SmallWord = System.Word; Integer = System.Longint; Word = System.Longint; const MaxInt = high(longint); type PByte = ^Byte; PWord = ^Word; PLongint = ^Longint; PSmallInt = ^SmallInt; PSmallWord = ^SmallWord; The ERRCODES unit ----------------- Contains constants for error codes, given by RUNERROR(), and the error_msg(code: Word): String, function that decyphers the error code. The DEBUG unit -------------- This module prints out the error code and the call stack in case of a run-time error. The stack is printed as follows: RunError #201 (range check error) Calls stack: SYSTEM.BOUND_ERROR [chk_fun.inp(21) at 0000000A] TEST.ASSN [TEST.PAS(61) at 00000015] TEST.TEST [TEST.PAS(82) at 0000001D] To use DEBUG, simply list it in the USES clause of the main program. +++0.33 Using the debug UNIT increases the .exe module size !!! Note: a call of a procedure with a NULL address is currently diagnosed as an arithmetic overflow, or (under the PMODE extender) caused a GP Fault or other traps ---0.33 The STRINGS unit ---------------- Corresponds to the STRING unit of BP. Also contains additional functions: Function hex (n: DWord ): String [12]; -- prints the argument in hex. Function whl (n: LongInt ): String [12]; -- prints the argument as a signed integer. Function uns (n: DWord ): String [12]; -- prints the argument as an unsigned integer. Function fix (x: Extended; pr: LongInt): String [15]; -- prints a real with fixed point. pr is the number of digits after the decimal point. Function align (str: String; width: LongInt): String; -- pads the argument up to width with spaces. spaces are on the right if width>0, and on the left otherwise. +++0.40 Additional functions defined in STRINGS: FUNCTION Dup_SI (CONST s: STRING; n: INTEGER): STRING; - copies the string "n" times FUNCTION Dup_æI (ß: CHAR; n: INTEGER): STRING; - creates the string by repeating character "c" "n" times. These functions are defined as operations: OVERLOAD * = Dup_SI; OVERLOAD * = Dup_CI; For example, you can write 'ab'*4 instead of 'abababab'. FUNCTION AppendStr (VAR dst: STRING; CONST src: STRING); - appends "src" at the end of "dst" FUNCTION AppendStrC (VAR dst: STRING; src: CHAR); - appends character "src" to the end of string "dst" These functions are defined as operations: OVERLOAD +:= = AppendStr; OVERLOAD +:= = AppendStrC; ---0.40 +++0.37 The MOUSE unit -------------- The MOUSE unit gives a Pascal program the access to the main mouse-support functions: * Determining presense/absense of the mouse * Mouse cursor positioning * Getting information about the position and the button states * Installing a mouse-driven Pascal interrupt handler. Variables: var ButtonCount: Word; During initilaization has value 0; after a call to InitMouse() this variable becomes either 0 (no mouse), or the number of available buttons (2 or 3). Procedures & Functions: Procedure InitMouse; Does the following: - Allocates a 4K stack for the user's mouse interrupt-driven procedure. - Determines the presence or absence of the mouse and sets the ButtonCount variable. - If mouse is present: - Enables the mouse cursor (calls ShowMouse) - Positions the cursor to the upper-left screen corner; - Reserves the callback address for the users' interrupt-driven mouse procedure (regardless of whether it will get installed or not). Procedure ShowMouse; When mouse is present, makes its cursor visible (INT 33 01). Procedure HideMouse; When mouse is present, makes its cursor invisible (INT 33 02). Procedure SetMousePos (X, Y: Integer); Positions the Mouse Cursor to the point (X,Y). Both coordinates start with 0. Procedure GetMouseInfo (var ButtonMask:Word; var X:Word; var Y:Word); Returns information about the position of the mouse cursor (X,Y) and the pressed buttons (ButtonMask). Procedure SetMouseHandler (Mask:Word; procedure Hnd (Mask, Buttons, X,Y, MovX,MovY:Word;)) Installs a Pascal interrupt-driven user's mouse handler (Hnd). The Mask parameter defines the classes of the events that call the handler; its format corresponds to the function INT 33 0C. When Hnd is called the Mask parameter contains the mask with the event type that occurred; Buttons contain the mask of the currently pressed buttons; X and Y contain the cursor's absolute position, and MovX and MovY contain the relative (signed) change of the last cursor position [negative numbers mean left or down; positive mean right or up]. These values are given in mouse position units; to convert to symbols, they need to be divided by 8. Using SetMouseHandler you can install several handlers with different masks without having to clear previous handlers. Procedure ClearMouseHandler; Removes the user's mouse interrupt handler. Procedure DoneMouse; Does the following: - removes (hides) the mouse cursor; - frees the callback address for the mouse interrupt handler; - frees the stack for the mouse interrupt handlers; - resets ButtonCount to 0. ---0.37 +++0.33 *******************VERSION 0.33 CHANGES******************************* This version of the compiler uses a new DOS extender (based on the PMODE package). It emulates the DOS32 functions that are necessary for run-time support of Pascal programs, which allows to use both the new extender as well as DOS32. The full list of the DOS32 functions, which are supported by the new extender is given at the end of this document The main difference of the new extender is that the 1st MB of the physical RAM is mapped to the 1st MB of the virtual memory while the program is loaded into higher addressed. This corresponds to the value _zero = 0. Thus, with the new extender it is possible to use absolute memory addressing: var vmem: array [1..25, 1..80] of word absolute $b8000; Because of this we could now implement some other BP7.0 constructs: mem [seg:offs] var x:type absolute seg:offs Here the effective address is computed as seg*16+offs. The Ptr (seg, offs) function works similarly. The seg(v) function still always returns 0. These new functions should substantially simplify the conversion of the programs that use absolute addressing. An example of using these functions can be found in file EXAMPLES\FLAME\FLAME1.PAS By default PLT uses the new extender (stub passtub.exe); the old one (DOS32) can be used by specifying the command line parameter -stub:dos32.exe or adding the line stub dos32.exe in the configuration file. The support for the DOS32 extender is likely to be entirely removed in the next version of PLT. Unless you discover some specific problems with the new extender (please report them if you do!), you should not use the DOS32 stub loader. --------- The following DOS32 functions are supported by the new extender: subfunctions of INT 31h: EE40h EE42h subfunctions of INT 21h: 39h, 3Ah, 3Bh, 3Ch, 3Dh, 3Fh, 40h 41h, 42h, 4300h, 4301h, 47h, 4Bh, 4Eh, 4Fh, 56h, 5Ah, 5Bh ********************************************************************** ---0.33 +++0.37 *******************VERSION 0.37 CHANGES******************************* New unit: MOUSE. No syntax changes; numerous errors in the PMODE-based extender and the compiler are corrected. ********************************************************************** ---0.37 +++0.40 *******************VERSION 0.40 CHANGES******************************* User defined operators added Iterated typed array constants added Extra Strings function added Many bugs fixed, code size of PMode extender reduced to 11KB. ********************************************************************** ---0.40 +++0.41 *******************VERSION 0.41 CHANGES******************************* New variable initialization within their declaration: TMT Pascal now allows variable declarations like: Var i,j: integer := n*n; k: integer := i+j; Expressions are computed and assigned whenever the block is entered. When more than one variable is initialized within the same expression, the expression is computed only once. Within the expressions you can use all the variables declared in the outside blocks, as well as the variables in the current block which declarations precede the expression. ALSO: Added directives ALIGN 2 ALIGN 4 in built-in assembler. The padding is done with the $90 byte (NOP). ********************************************************************** ---0.41 +++1.00 *******************VERSION 1.00 CHANGES******************************* A number of bugs fixed; the quality of code generated improved; full on-line documentation added. ---1.00 Possible problems during usage ------------------------------ Programs, generated by TMT Pascal, abend under OS/2. Correction: - ensure that in "memory settings" EMS_MEMORY_LIMIT is not equal to 0. +++0.33 Run-time errors codes -------- ------ ----- 1 Invalid function number 2 File not found 3 Path not found 4 Too many open files 5 File access denied 6 Invalid file handle 12 Invalid file access_code 15 Invalid drive number 16 Cannot remove current directory 17 Cannot rename across drives 18 No more files 100 Disk read error 101 Disk write error 102 File not assigned 103 File not open 104 File not open for input 105 File not open for output 106 Invalid numeric format 150 Disk is write protected 151 Bad drive request struct length 152 Drive not ready 154 CRC error in data 156 Disk seek error 157 Unknown media type 158 Sector not found 159 Printer out of paper 160 Device write fault 161 Device read fault 162 Hardware failure 200 Division by zero 201 Range check error 202 Stack overflow error 203 Heap overflow error 204 Invalid pointer operation 205 Floating point overflow 206 Floating point underflow 207 Invalid floating point operation 208 Overlay manager not installed 209 Overlay file read error 210 Object not initialized 211 Call to abstract method 212 Stream registration error 213 Collection index_out of range 214 Collection overflow error 215 Arithmetic overflow error 216 General protection fault 217 Invalid operation_code 300 File io error 301 Nonmatched array bounds 302 Non local procedure pointer 303 Procedure pointer out of scope 304 Function not implemented 305 Breakpoint error 306 Break by ctrl/C 307 Break by ctrl break 308 Break by other process 309 No floating point coprocessor ---0.33