Revista CD Expert 8/Revista CD Expert nº 08 CD1.iso/Utilitarios/Programacao/Pacific C for DOS/HELP/PPD.HLP

Text File | 1995-11-14 | 13.9 KB | 317 lines

INDEX 7 106 PPD
1390 88 Compiler optimizations
7777 106 Editor Searching
6183 11 Heap size
30 28 Memory models and chip types
13403 20 PPD command line options
11807 27 String search
6792 21 User defined commands
%Memory models and chip types
Available processor types are:

8086 or 8088    This selection will produce code that will run on any
                8086 family processor.

80186, 80188 or 80286
                This chip selection will allow the compiler to generate
                code using instructions only present on these processors
                (or on later processors like the 80386).

Small           This memory model allows one 64K code segment and one
                64K data and stack segment.  More data may be accessed
                via far pointers and variables.

Large           The large model allows multiple code and data segments.
                The only limit on code or data is memory size.  Due to
                the 8086 architecture, the stack is still limited to 64K.

Hardware (8087) floating point
                This toggle will tell the compiler whether to generate
                inline 8087 code or calls to the floating point library.
                8087 code will not run unless an 8087, 80287 or 80387
                co-processor is present on an 80486 or Pentium processor
                is being used.  Code using the floating point library
                calls will still take advantage of an 8087 because the
                floating point library auto-detects the presence of a
                co-processor.

%Compiler optimizations
An "optimization" is a transformation performed by the compiler on
a program to make it smaller and/or faster.

The compiler performs many optimizations, some automatically, some
only when requested.  The Optimization dialog box allows selection
of optimizations to be performed when compiling:

Global:         Global optimization is performed one function at
                a time and aims to optimize the use of registers
                within whole functions.  When used, this allocates
                variables to registers based on analysis of which
                variables will benefit most. This optimization can
                be quite time consuming on large functions.  It is
                possible for the code generator to run out of
                memory when performing global optimization. The
                function size should be reduced if this occurs.

Peephole:       Peephole optimization is a process of examining
                generated assembler code and "improving" it. This
                includes redundant load/store optimization, branch
                optimization, and substitution of faster or shorter
                instructions.  Whole functions are optimized at a
                time, and it is possible for the optimizer to run
                out of memory on large functions.

Assembler:      The assembler attempts to turn long branches or
                jumps into short branches when this is used. This
                does not use extra memory, but can increase the
                compile time.

Alignment:      The compiler includes an optional optimization pass
                which aligns function entry points, code following
                JMP instructions and labels within the code on
                specified memory boundaries.  Alignment is userful
                when producing code to for newer 80x86 processors
                such as the 80386, 80486 and Pentium.  This pass
                will be skipped if no alignments are specified.

                Each class of alignment may be individually enabled
                or disabled based on values entered in "Optimization"
                dialog.  A value of 0 means that a particular form
                of alignment is disabled.  Other valid alignments are
                2, 4, 8 and 16.  Alignments of 2 are useful on 8086,
                80186, 80286 and 80386SX processors.  4 is optimal
                on the 80386, use 16 on the 80486 and Pentium.

                When selecting alignment values bear in mind that
                you are trading off code size for relatively minor
                improvements in program performance considering that
                the newer processors where large alignments help are
                much much faster than earlier processors anyway.

                For the best size/speed tradeoffs, suggested values
                are as follows:

                8088, 80C188          no alignment optimization
                8086/186/286/386SX    function:2,  jump:2, label:2
                80386DX               function:4,  jump:4, label:4
                80486, Pentium        function:16, jump:4, label:4

                The three alignment optimizations available are:

Function:       This value specifies the memory alignment on which
                C function entry points will be located.  This has
                the least effect on program size of all alignments
                and can improve performance of small functions on
                the 80486 and Pentium.  Use 2 on 80286 processors,
                4 on 80386 and 16 on 80486 and Pentium.

Jump:           This value specifies a boundary on which code after
                JMP instructions will be placed.  JMP alignment can
                improve the performance of while(), for() switch()
                and goto constructs, but has considerably more impact
                on program size than function entry point alignment.
                For 80386, 80486 and Pentium an alignment of 4 is a
                good tradeoff.

Label:          This value specifies the alignment which will be used
                for compiler generated labels starting with "l" and
                "L".  This optimization improves the performance of
                if() constructs, switch() cases and other code which
                performs conditional branches.  Note that this may
                make code run slower where a conditional branch is not
                taken, as NOP instructions will be used to pad before
                each label.  Generally, values of 0, 2 or 4 should be
                used for label alignment.  16 will cause a sustantial
                code size increase for very speed improvement

%Heap size
The "heap size" value is placed in the .EXE file header to tell DOS
how much heap and stack space should initially be allocated to the
program.  With Pacific C, it is used to limit the size of "near"
data and heap, there is no "far" heap as such because the Pacific C
memory allocation routines use MS-DOS system calls to obtain memory.

The heap size should be entered as a hexadecimal value between 0 and
10000 (hex) inclusive.  The default value is 0, which tells Pacific C
to use a default heap size of 64K bytes.  If a value larger than 64K
is specified, it will not be accepted.

%User defined commands
The "User command setup" dialog allows four user defined external
commands to be added to the Utility menu.  Each user defined command
entry consists of two strings; the menu entry which will be displayed
in the Utility menu, and the command line which will be executed by
DOS when the menu item is selected.  Typically the user commands will
be used to interface to external debuggers and editors.

A number of macros may be used within user defined commands to allow
substitution of useful filenames.  The available macros are:

        $(EDIT)       Current PPD edit file
        $(OUTFILE)    Current output file name
        $(CWD)        Current working directory
        $(LIB)        Compiler LIB directory path
        $(INC)        Compiler INCLUDE directory path
        $(PROJ)       Name of current project file

For example, a user defined command which passes the current PPD
edit file to an external editor called EDIT would be:

        EDIT $(EDIT)
%Editor Searching
The editor allows search and search/replace forwards or backwards, using
either simple strings, or "regular expressions".  The search/replace dialog
can be accessed via several paths:

1.      by using the WordStar style commands Ctrl-Q A and Ctrl-Q F.
2.      by clicking the "Search" button in the editor window.
3.      by pressing F2 in the editor window.
4.      by the "Search ..." or "Replace ..." items in the "Edit" menu.

The search/replace dialog allows both search and replace operations to be
selected, starting from any of the search/replace commands.  The settings
in the dialog will default to appropriate values based on the previous
operation and the method used to activate the dialog.  Selection between
search and replace operations is performed using the "Replace - F3" check
box.  If performing a replace operation, the check box "Global - F5" allows
an interactive global search/replace to be performed.  The "Replace all - F6"
check box allows all instances of the search string to be replaced without
asking.  You can select between case sensitive and case insensitive searching
using the "Case sensitive - F4" check box.

In addition to the search/replace commands which display the dialog, the
editor also allows the previous search/replace operation to be repeated
in either a forward or backward direction.  This can be achieved using the
following key or mouse strokes:

Search for NEXT:          Ctrl-L or Shift-F2 or click the NEXT button.
Search for PREVIOUS:      Ctrl-L or Ctrl-F2 or click the PREV button.

REGULAR EXPRESSION MATCHING

Regular expressions are a generalization of "wild card" matching.  A
regular expression can include literal text (e.g. "abcdef") or various
special characters that are interepreted specially when comparing the
search string with the text in the editor.  In brief, the special
characters are:

^       Match start of line
$       Match end of line
?       Match any character
*       Repeat previous expression
\( \)   Group a sequence
[ ]     Specify a range or set of characters to match
\       Escape a special character

In more detail:

^       Use at the beginning of a search string to anchor the search
        to the beginning of the line, e.g.

           ^#define

        will match #define only if it is at the beginning of a line.

$       Matches the end of a line, e.g.

           0);$

        will match "0);" only at the end of a line.

?       matches any character, e.g.

          ??xyz?

        would match any of:

          "12xyzb", "..xyz1" or "AAxyzA"

*       means any number of repetitions of the previous regular
        expression.  The previous regular expression is either a
        single character, or a parenthesized expression or range
        as below. For example:

           ab*

        will match the letter 'a' followed by zero or more letters 'b'.

\(  \)  These two character pairs enclose a regular expression that
        can be the subject of the * repetition character, e.g

           ab\(ab\)*

        will match one or more sequences of "ab", e.g. "ababab". Note
        that the first "ab" must be matched with "ab", and the sequence
        inside the parentheses can match zero or more sequences of "ab".

[   ]   These brackets are used to enclose a character range or set.
        Within the brackets can be any number of single characters,
        or pairs of characters separated by a dash. For example:

          [ABCDEF0-9]

        will match any upper case alphabetic from A to F, or any digit.
        If it is desired to include a dash in the set, make it the first
        or last character. Similarly, the right bracket (]) must be first
        if it is to be part of the set. For example:

          []a-f-]

        will match any lower case alphabetic, ] or -.

\       Escape any special character. Precede any of ?, *, $, ^ or [
        to force them to be treated literally. For example:

          \*\**

        will match one or more asterisks in sequence.
%String search
The string search dialog is used to access a "grep" like utility which
will perform a search of one or more files for a specified target string.
The search string may consist of any sequence of characters, including
spaces.  The default behaviour is to perform a case insensitive search
where lower case and upper case characters are treated as equivalent.
If the "Case sensitive - F4" check box is selected, a case sensitive
search may be performed.

The list of files to search may be specified either by typing a list
of wildcards in the "File list:" edit line, or by selecting the check
box: "Current project - F6".  The file list entered may consist of one
or more file names or wildcards.  If the same file is selected by more
than one wildcard, the file will only be searched once.  If "Current
project" is selected, all source files in the current PPD project will be
searched for the target string.  If you wish to search a user specified
file list, select the "File list - F5" check box.

The results of the search will be displayed in the "String search:" window.
You may move between entries in this window using the mouse, window scroll
bar, or the arrow/cursor keys on the keypad.  To select a search result,
double click the mouse in its entry, or press enter when the desired line
is highlighted.  When a selection is made, the selected file will be loaded
into the PPD editor with the cursor and search highlight positioned on
the selected text.

The "Search F2" button in the "String search:" window may be used to search
again without using the menu.
%PPD command line options
Pacific C is invoked by the PPD command. Options to the
PPD command are:

        /screen:25
        /screen:28
        /screen:43
        /screen:50

These options set the DOS screen size to the specified number
of lines. A VGA or better display is needed for 50 lines, an
EGA or better for the others.

PPD will also take a file name, which may be a project file
name or a source file name. For example:

ppd /screen:43 testing

will start up PPD in 43 line mode, and look for a file TESTING.PRJ
or a file TESTING.C. A project file will be loaded, and a source
file will be loaded into the editor.