Porting C-Source to OS/400 - Revision history

PoC: /* Weblinks */ more precise

2026-05-17T19:05:17Z

Weblinks: more precise

← Older revision		Revision as of 21:05, 17 May 2026
Line 56:		Line 56:

	== Weblinks ==		== Weblinks ==
	* [https://github.com/greenfive/as400_c/ AS/400 C] programming related files, especially ''stdint.h''		* [https://github.com/greenfive/as400_c/ AS/400 C] programming related files, especially ''stdint.h'', GitHub

	== Footnotes ==		== Footnotes ==

PoC: +link

2026-05-17T19:04:55Z

+link

← Older revision		Revision as of 21:04, 17 May 2026
Line 54:		Line 54:
	If all object modules have been created, you can then link them to a ''*PGM'' object as follows:		If all object modules have been created, you can then link them to a ''*PGM'' object as follows:
	CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)		CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)

			== Weblinks ==
			* [https://github.com/greenfive/as400_c/ AS/400 C] programming related files, especially ''stdint.h''

	== Footnotes ==		== Footnotes ==

PoC: More types

2026-05-10T23:38:44Z

More types

← Older revision		Revision as of 01:38, 11 May 2026
Line 6:		Line 6:
	These are definitions for integers, being used often. This can be put e. g. into a separate header file:		These are definitions for integers, being used often. This can be put e. g. into a separate header file:
	<syntaxhighlight lang="c">		<syntaxhighlight lang="c">
	#ifndef __stdint_h		#ifndef __stdint_h
	typedef unsigned long long int uint64_t;		typedef unsigned long long int uint64_t;
	typedef unsigned int uint32_t;		typedef unsigned int uint32_t;
	typedef unsigned short int uint16_t;		typedef unsigned short int uint16_t;
	typedef unsigned char uint8_t;		typedef unsigned char uint8_t;
	#define __stdint_h 1		typedef long long int int64_t;
	#endif		typedef int int32_t;
			typedef short int int16_t;
			typedef char int8_t;
			#define __stdint_h 1
			#endif
	</syntaxhighlight>		</syntaxhighlight>
	<blockquote>'''Note:''' The IMPI (CISC) C compiler in V3R2 and earlier doesn't know <tt>long long</tt>, it's 32 bits only.</blockquote>		<blockquote>'''Note:''' The IMPI (CISC) C compiler in V3R2 and earlier doesn't know <tt>long long</tt>, it's 32 bits only.</blockquote>

Greenfive: fixed typo

2026-04-28T21:55:17Z

fixed typo

← Older revision		Revision as of 23:55, 28 April 2026
Line 17:		Line 17:

	== Preparations for QSYS.LIB ==		== Preparations for QSYS.LIB ==
	Preparations are mostly ~~one~~ on UNIX/Linux side. This is required to make the older V4R5 compiler happy. Newer compilers might be more forgiving.		Preparations are mostly done on UNIX/Linux side. This is required to make the older V4R5 compiler happy. Newer compilers might be more forgiving.

	# Convert to native line endings for your platform, if the source files don't match.		# Convert to native line endings for your platform, if the source files don't match.

PoC: header

2025-12-31T15:38:04Z

header

← Older revision		Revision as of 17:38, 31 December 2025
Line 4:		Line 4:
	Most often, the real challenge is to manually put together a ''config.h'' file. Since there's no capable UNIX shell environment, <code>./configure</code> is no option.<ref>With V4R5, IBM claims to have implemented the C90 standard in Compiler and C-Library. I did not yet find a readymade ''config.h'' with all statements needed for a C90 environment.</ref>		Most often, the real challenge is to manually put together a ''config.h'' file. Since there's no capable UNIX shell environment, <code>./configure</code> is no option.<ref>With V4R5, IBM claims to have implemented the C90 standard in Compiler and C-Library. I did not yet find a readymade ''config.h'' with all statements needed for a C90 environment.</ref>

	These are definitions for integers, being used often.		These are definitions for integers, being used often. This can be put e. g. into a separate header file:
	<syntaxhighlight lang="c">		<syntaxhighlight lang="c">
	typedef unsigned long long int uint64_t;		#ifndef __stdint_h
	typedef unsigned int uint32_t;		typedef unsigned long long int uint64_t;
	typedef unsigned short int uint16_t;		typedef unsigned int uint32_t;
			typedef unsigned short int uint16_t;
			typedef unsigned char uint8_t;
			#define __stdint_h 1
			#endif
	</syntaxhighlight>		</syntaxhighlight>
	<blockquote>'''Note:''' The IMPI (CISC) C compiler in V3R2 and earlier doesn't know <tt>long long</tt>, it's 32 bits only.</blockquote>		<blockquote>'''Note:''' The IMPI (CISC) C compiler in V3R2 and earlier doesn't know <tt>long long</tt>, it's 32 bits only.</blockquote>

PoC: Reworked

2025-01-19T22:35:42Z

Reworked

← Older revision		Revision as of 00:35, 20 January 2025
Line 3:		Line 3:

	Most often, the real challenge is to manually put together a ''config.h'' file. Since there's no capable UNIX shell environment, <code>./configure</code> is no option.<ref>With V4R5, IBM claims to have implemented the C90 standard in Compiler and C-Library. I did not yet find a readymade ''config.h'' with all statements needed for a C90 environment.</ref>		Most often, the real challenge is to manually put together a ''config.h'' file. Since there's no capable UNIX shell environment, <code>./configure</code> is no option.<ref>With V4R5, IBM claims to have implemented the C90 standard in Compiler and C-Library. I did not yet find a readymade ''config.h'' with all statements needed for a C90 environment.</ref>

			These are definitions for integers, being used often.
			<syntaxhighlight lang="c">
			typedef unsigned long long int uint64_t;
			typedef unsigned int uint32_t;
			typedef unsigned short int uint16_t;
			</syntaxhighlight>
			<blockquote>'''Note:''' The IMPI (CISC) C compiler in V3R2 and earlier doesn't know <tt>long long</tt>, it's 32 bits only.</blockquote>

	== Preparations for QSYS.LIB ==		== Preparations for QSYS.LIB ==
	Preparations are mostly one on UNIX/Linux side.		Preparations are mostly one on UNIX/Linux side. This is required to make the older V4R5 compiler happy. Newer compilers might be more forgiving.

			# Convert to native line endings for your platform, if the source files don't match.
	# Convert Tabs to spaces:<ref>Tabs appear in SEU as only one blank. This makes source indentation hardly visible.</ref><br><code>for FILE in .c .h; do cp ${FILE} ${FILE}~; expand -i -t4 < ${FILE}~ > ${FILE}; done</code>		# Convert Tabs to spaces:<ref>Tabs appear in SEU as only one blank. This makes source indentation hardly visible.</ref><br><code>for FILE in .c .h; do cp ${FILE} ${FILE}~; expand -i -t4 < ${FILE}~ > ${FILE}; done</code>
	# Find out longest line in all C- and H-Files.<br><code>wc -L *.c \|sort -rn \|head</code>		# Find out longest line in all C- and H-Files.<br><code>wc -L *.c \|sort -rn \|head</code>
	# Eliminate too long lines by wrapping. This is easily done with <code>set colorcolumn=80</code> in ''vim'', and scrolling through the files with a wide window. <ref>Wider than 80 chars, that is.</ref>		# Eliminate too long lines by wrapping. This is easily done with <code>set colorcolumn=80</code> in ''vim'', and scrolling through the files with a wide window. <ref>Wider than 80 chars, that is.</ref>
			# Find C++-style comments and convert to proper C comments.
			# Find function definitions which should be inlined, remove the word "inline" from the definition, and add a hint for the preprocessor/compiler that a function should be inlined before the function definition: <code>#pragma inline (''functionname'')</code>
			# Find and remove text following <code>#endif</code> preprocessor statements in the same line (thought as comment).
	# Create (at least) '''two''' appropriate SRCPF with matching maximum line lengths through the RCDLEN parameter for CRTSRCPF:		# Create (at least) '''two''' appropriate SRCPF with matching maximum line lengths through the RCDLEN parameter for CRTSRCPF:
	## One (or more) for C-Files (''may'' be named ''QCSRC''),<ref>It could be clever to create one PF per subdirectory.</ref>		## One (or more) for C-Files (''may'' be named ''QCSRC''),<ref>It could be clever to create one PF per subdirectory.</ref>
Line 15:		Line 27:
	# Upload.		# Upload.
	## If the file names are already shorter than 10 Characters (excluding file extension), just upload to a SRC PF:<br><code>for FILE in *.c; do echo -n "put ${FILE} MYLIB/QCSRC."; basename ${FILE} .c; done \|ftp as400</code><ref>Automatic login and switching to ASCII mode is done though a ''~/.netrc''-file.</ref><ref>FTP ASCII takes care to convert ASCII to EBCDIC.</ref>.		## If the file names are already shorter than 10 Characters (excluding file extension), just upload to a SRC PF:<br><code>for FILE in *.c; do echo -n "put ${FILE} MYLIB/QCSRC."; basename ${FILE} .c; done \|ftp as400</code><ref>Automatic login and switching to ASCII mode is done though a ''~/.netrc''-file.</ref><ref>FTP ASCII takes care to convert ASCII to EBCDIC.</ref>.
	## If the file names are longer, you'll have considerable fun to come up with shorter names~~, possibly~~ ''grep'' through and replace in all text files if you needed to shorten a ''.h'' name, ~~manually~~ put together a ~~textfile~~ with ~~approopriate~~ ''put'' commands, and hope for the best.		## If the file names are longer, you'll have considerable fun to come up with shorter names. Possibly ''grep'' through, and replace names in all text files if you needed to shorten a ''.h'' name. Finally, put together a text file with appropriate ''put'' commands, and hope for the best.
	# Change file type to ''C'':<br><code>for FILE in *.c; do echo "quote rcmd CHGPFM FILE(MYLIB/QCSRC) MBR(`basename ${FILE} .c`) SRCTYPE(C)"; done \|ftp as400</code><ref>This is a good example how to exploit the ftp server's ability to call CL commands as a replacement for ''rcmd''.</ref>		# Change file type to ''C'':<br><code>for FILE in *.c; do echo "quote rcmd CHGPFM FILE(MYLIB/QCSRC) MBR($(basename ${FILE} .c)) SRCTYPE(C)"; done \|ftp as400</code><ref>This is a good example how to exploit the ftp server's ability to call CL commands as a replacement for ''rcmd''.</ref> Note: You'll need to do this for header files, also!

	The precompiler searches private includes in ''*LIBL/H'', without the ''.h''-Extension.		The precompiler searches private includes in ''*LIBL/H'', without the ''.h''-Extension.
Line 22:		Line 34:
	== Compile Cycle ==		== Compile Cycle ==
	Compile object files as follows:		Compile object files as follows:
	CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)<~~ref~~>~~Possibly add ''~~SYSIFCOPT(*IFS64IO)'' ~~if large file support is desired.~~</~~ref~~>		CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)
			* Add <code>SYSIFCOPT(IFSIO IFS64IO)</code> to the compiler parameters if the code contains <code>fopen()</code> calls.
			* If you added <code>#pragma inline (''functionname'')</code> to the code, add <code>INLINE(*ON)</code> to the compiler parameters

	If compilation fails, output will go into a spooled file in ~~''QEZJOBLOG''~~ queue. This is the most tedious task. Compile, check output, fix problems, ~~repeat~~.		If compilation fails, output will go into a spooled file in the configured output queue. This is the most tedious task.
			# Compile,
			# check output,
			# fix problems,
			# <code>goto 1</code>.

	~~Link~~ object ~~files~~ to a ~~program~~ as follows:		If all object modules have been created, you can then link them to a ''*PGM'' object as follows:
	CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)		CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)

PoC: User error, thus no article.

2021-05-18T16:23:43Z

User error, thus no article.

← Older revision		Revision as of 18:23, 18 May 2021
Line 28:		Line 28:
	Link object files to a program as follows:		Link object files to a program as follows:
	CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)		CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)

	~~== See also ==~~
	* [[Coded Character Set issues with FTP]]

	== Footnotes ==		== Footnotes ==

PoC: +Link

2021-05-18T15:02:16Z

+Link

← Older revision		Revision as of 17:02, 18 May 2021
Line 28:		Line 28:
	Link object files to a program as follows:		Link object files to a program as follows:
	CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)		CRTPGM PGM(MYLIB/MYPG) MODULE(MYLIB/ALL) ENTMOD(PGM) ACTGRP(NEW) DETAIL(BASIC)

			== See also ==
			* [[Coded Character Set issues with FTP]]

	== Footnotes ==		== Footnotes ==

PoC: Added heading

2020-07-18T10:43:11Z

Added heading

← Older revision		Revision as of 12:43, 18 July 2020
Line 20:		Line 20:
	The precompiler searches private includes in ''*LIBL/H'', without the ''.h''-Extension.		The precompiler searches private includes in ''*LIBL/H'', without the ''.h''-Extension.

			== Compile Cycle ==
	Compile object files as follows:		Compile object files as follows:
	CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)<ref>Possibly add ''SYSIFCOPT(*IFS64IO)'' if large file support is desired.</ref>		CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)<ref>Possibly add ''SYSIFCOPT(*IFS64IO)'' if large file support is desired.</ref>

PoC: More precise

2020-07-16T09:54:37Z

More precise

← Older revision		Revision as of 11:54, 16 July 2020
Line 23:		Line 23:
	CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)<ref>Possibly add ''SYSIFCOPT(*IFS64IO)'' if large file support is desired.</ref>		CRTCMOD MODULE(MYLIB/MAIN) SRCFILE(MYLIB/QCSRC) DEFINE(DEF1 DEF2=1) OPTION(LOGMSG) OUTPUT(PRINT)<ref>Possibly add ''SYSIFCOPT(*IFS64IO)'' if large file support is desired.</ref>

	If compilation fails, output will go into a spooled file in ''QEZJOBLOG'' queue.		If compilation fails, output will go into a spooled file in ''QEZJOBLOG'' queue. This is the most tedious task. Compile, check output, fix problems, repeat.

	Link object files to a program as follows:		Link object files to a program as follows: