Sign in Join
FPDoc Onlineexperiment

Project management and compiler options

Project management in Pascal is much easier than with C. The compiler knows from the source which units, sources etc. it needs. So the Free PascalIDE does not need a full featured project manager like some C development environments offer. Nevertheless there are some settings in the IDE which apply to projects.

The primary file

Without a primary file the IDE compiles/runs the source of the active window when a program is started. If a primary file is specified, the IDE always compiles/runs this source, even if another source window is active. With the menu item a file dialog can be opened where the primary file can be selected. Only the menu item compiles the active window regardless. This is useful if a large project is being edited, and only the syntax of the current source should be checked.

The menu item restores the default behavior of the IDE, i.e. the ’compile’ and ’run’ commands apply to the active window.

The directory dialog

In the directory dialog, the directories can be specified where the compiler should look for units, libraries, object files. It also says where the output files should be stored. Multiple directories (except for the output directory) can be entered, separated by semicolons. The directories dialog is shown in figure [fig:odirs].

The following directories can be specified:

EXE & PPU directories

Specifies where the compiled units and executables will go. ( on the command line.)

Object directories

Specifies where the compiler looks for external object files. ( on the command line.)

Library directories

Specifies where the compiler (more exactly, the linker) looks for external libraries. ( on the command line.)

Include directories

Specifies where the compiler will look for include files, included with the {$i } directive. ( or on the command line.)

Unit directories

Specifies where the compiler will look for compiled units. The compiler always looks first in the current directory, and also in some standard directories. ( on the command line.)

Remark: For the search path an asterisk (*) as the last directory will include all subdirectories one level deep.

Path substitution variables such as $fpctarget and $fpcversion can be used (section Compiler configuration).

The target operating system

The menu item allows specification of the target operating system for which the sources will be compiled. Changing the target doesn’t affect any compiler switches or directories. It does affect some defines defined by the compiler. The settings here correspond to the option on the command line . A sample compilation target dialog is shown in figure [fig:target]: the actual dialog will show only those targets that the IDE actually supports.

The following targets can be set (the list depends on the platform for which the IDE was compiled):

Dos (go32v1)

This switch will disappear in time as this target is no longer being maintained.

Dos (go32v2)

Compile for DOS, using version 2 of the Go32 extender.

FreeBSD

Compile for FreeBSD.

Linux

Compile for Linux.

OS/2

Compile for OS/2 (using the EMX extender).

Windows

Compile for Windows.

The currently selected target operating system is shown in the menu item in the menu. Initially, this will be set to the operating system for which the IDE was compiled.

Compiler options

The menu allow the setting of options that affect the compilers behavior. When this menu item is chosen, a dialog pops up that displays several tabs.

There are six tabs:

Syntax

Here options can be set that affect the various syntax aspects of the code. They correspond mostly to the -S option on the command line (section Compiler configuration).

Code generation

These options control the generated code; they are mostly concerned with the -C and -X command line options.

Verbose

These set the verbosity of the compiler when compiling. The messages of the compiler are shown in the compiler messages window (can be called with ).

Browser

Options concerning the generated browser information. Browser information needs to be generated for the symbol browser to work.

Assembler

Options concerning the reading of assembler blocks (-R on the command line) and the generated assembler (-A on the command line)

Processor

Here the target processor can be selected.

On each tab page, there are two entry boxes: the first for Conditional defines and the second for additional compiler arguments. The symbols, and arguments, should be separated with semi-colons.

The syntax tab of the compiler options dialog is shown in figure [fig:ocompa].

In the syntax options dialog, the following options can be set:

Stop after first error

when checked, the compiler stops after the first error. Normally the compiler continues compiling till a fatal error is reached. ( on the command line)

Allow label and goto

Allow the use of label declarations and goto statements ( on the command line).

Enable macros

Allow the use of macros ().

Allow inline

Allow the use of inlined functions ( on the command line).

Include assertion code

Include Assert statements in the code.

Load kylix compat. unit

Load the Kylix compatibility unit.

Allow STATIC in objects

Allow the Static modifier for object methods ( on the command line)

C-like operators

Allows the use of some extended operators such as +=, -= etc. ( on the command line).

Compiler mode

select the appropriate compiler mode:

Free Pascal Dialect

The default Free Pascalcompiler mode (FPC).

Object pascal extensions on

Enables the use of classes and exceptions ( on the command line).

Turbo pascal compatible

Try to be more Turbo Pascalcompatible ( on the command line).

Delphi compatible

Try to be more Delphicompatible ( on the command line).

Macintosh Pascal dialect

Try to be Macintosh pascal compatible.

The code generation tab of the compiler options dialog is shown in figure [fig:ocompb].

In the code generation dialog, the following options can be set:

Run-time checks

Controls what run-time checking code is generated. If such a check fails, a run-time error is generated. The following checking code can be generated:

Range checking

Checks the results of enumeration and subset type operations ( command line option).

Stack checking

Checks whether the stack limit is not reached ( command line option).

I/O checking

Checks the result of IO operations ( command line option).

Integer overflow checking

Checks the result of integer operations ( command line option).

Object method call checking

Check the validity of the method pointer prior to calling it.

Position independent code

Generate PIC code.

Create smartlinkable units

Create smartlinkable units.

Optimizations

What optimizations should be used when compiling:

Generate faster code

Corresponds to the -OG command line option.

Generate smaller code

Corresponds to the -Og command line option.

More information on these switches can be found in section Compiler configuration.

The processor tab of the compiler options dialog is shown in figure [fig:ocompf].

In the processor dialog, the target processor can be set. The compiler can use different optimizations for different processors.

The verbose tab of the compiler options dialog is shown in figure [fig:ocompc].

In this dialog, the following verbosity options can be set (on the command line: ):

Warnings

Generate warnings. Corresponds to -vw on the command line.

Notes

Generate notes. Corresponds to -vn on the command line.

Hints

Generate hints. Corresponds to -vh on the command line.

General info

Generate general information. Corresponds to -vi on the command line.

User,tried info

Generate information on used and tried files. Corresponds to -vut on the command line.

All

Switch on full verbosity. Corresponds to -va on the command line.

Show all procedures if error

If an error using overloaded procedure occurs, show all procedures. Corresponds to -vb on the command line.

The browser tab of the compiler options dialog is shown in figure [fig:ocompd].

In this dialog, the browser options can be set:

No browser

(default) No browser information is generated by the compiler.

Only global browser

Browser information is generated for global symbols only, i.e. symbols defined not in a procedure or function (-b on the command line)

Local and global browser

Browser information is generated for all symbols, i.e. also for symbols that are defined in procedures or functions (-bl on the command line)

Remark: If no browser information is generated, the symbol browser of the IDE will not work.

The assembler tab of the compiler options dialog is shown in figure [fig:ocompe]. The actual dialog may vary, as it depends on the target CPU the IDE was compiled for.

In this dialog, the assembler reader and writer options can be set:

Assembler reader

This permits setting the style of the assembler blocks in the sources:

AT&T assembler

The assembler is written in AT&T style assembler (-Ratt on the command line).

Intel style assembler

The assembler is written in Intel style assembler blocks (-Rintel on the command line).

remark that this option is global, but locally the assembler style can be changed with compiler directives.

Assembler info

When writing assembler files, this option decides which extra information is written to the assembler file in comments:

List source

The source lines are written to the assembler files together with the generated assembler (-al on the command line).

List register allocation

The compiler’s internal register allocation/deallocation information is written to the assembler file (-ar on the command line).

List temp allocation

The temporary register allocation/deallocation is written to the assembler file. (-at on the command line).

List node allocation

The node allocation/deallocation is written to the assembler file. (-an on the command line).

use pipe with assembler

use a pipe on unix systems when feeding the assembler code to an external assembler.

The latter three of these options are mainly useful for debugging the compiler itself, it should rarely be necessary to use these.

Assembler output

This option tells the compiler what assembler output should be generated.

Use default output

This depends on the target.

Use GNU as

Assemble using GNUas (-Aas on the command line).

Use NASM coff

Produce NASM coff assembler (go32v2, -Anasmcoff on the command line)

Use NASM elf

Produce NASM elf assembler (Linux, -Anasmelf on the command line).

Use NASM obj

Produce NASM obj assembler (-Anasmobj on the command line).

Use MASM

Produce MASM (Microsoft assembler) assembler (-Amasm on the command line).

Use TASM

Produce TASM (Turbo Assembler) assembler (-Atasm on the command line).

Use coff

Write binary coff files directly using the internal assembler (go32v2, -Acoff on the command line).

Use pecoff

Write binary pecoff files files directly using the internal writer. (Win32)

Linker options

The linker options can be set in the menu . It permits the specification of how libraries and units are linked, and how the linker should be called. The linker options dialog is shown in figure [fig:olinker].

The following options can be set:

Call linker after

If this option is set then a script is written which calls the linker. This corresponds to the s option on the command line ().

Only link to static library

Only use static libraries.

Preferred library type

With this option, the type of library to be linked in can be set:

Target default

This depends on the platform.

Dynamic libraries

Tries to link in units in dynamic libraries. (option -XD on the command line.)

Static libraries

Tries to link in units in static libraries. (option -XS on the command line.)

Smart libraries

Tries to link in units in smart-linked libraries. (option -XX on the command line.)

Memory sizes

The memory sizes dialog (reachable via ) permits the entry of the memory sizes for the project. The memory sizes dialog is shown in figure [fig:omemsize].

The following sizes can be entered:

Stack size

Sets the size of the stack in bytes (option -Cs on the command line). This size may be ignored on some systems.

Heap size

Sets the size of the heap in bytes; (option -Ch on the command line). Note that the heap grows dynamically as much as the OS allows.

Debug options

In the debug options dialog (reachable via ), some options for inclusion of debug information in the binary can be set; it is also possible to add additional compiler options in this dialog. The debug options dialog is shown in figure [fig:odebug].

The following options can be set:

Debugging information

tells the compiler which debug information should be compiled in. One of the following options can be chosen:

Strip all debug symbols from executable

Will strip all debug and symbol information from the binary. (option -Xs on the command line).

Skip debug information generation

Do not generate debug information at all.

Generate debug symbol information

Include debug information in the binary (option -g on the command line). Please note that no debug information for units in the Run-Time Library will be included, unless a version of the RTL compiled with debug information is available. Only units specific to the current project will have debug information included.

Generate also backtrace line information

Will compile with debug information, and will additionally include the lineinfo unit in the binary, so that in case of an error the backtrace will contain the file names and line numbers of procedures in the call-stack. (Option -gl on the command line.)

Generate valgrind compatible debug info

Generate debug information that can be read with valgrind (a memory checking tool).

Profiling switches

Tells the compiler whether or not profile code should be included in the binary.

No profile information

Has no effect, as it is the default.

Generate Profile code for gprof

If checked, profiling code is included in the binary (option -p on the command line).

Use another TTY for Debuggee

An attempt will be made to redirect the output of the program being debugged to another window (terminal), whose file name should be entered here.

The switches mode

The IDE allows saving a set of compiler settings under a common name. It provides 3 names under which the switches can be saved:

Normal

For normal (fast) compilation.

Debug

For debugging; intended to set most debug switches on. Also useful for setting conditional defines that e.g. allow including some debug code.

Release

For a compile of the program as it should be released, debug information should be off, the binary should be stripped, and optimizations should be used.

Selecting one of these modes will load the compiler options as they were saved the last time the selected mode was active, i.e. it doesn’t specifically set or unset options.

When setting and saving compiler options, be sure to select the correct switch mode first; it makes little sense to set debug options while the release switch is active. The switches mode dialog is shown in figure [fig:oswitch].

An experiment, built from development snapshots. Not the official Free Pascal or Lazarus documentation.