# [PATCH 1/4] Support builds when sys/param.h is missing 4 messages from 2012-12-14 to 2012-12-15. Participants: David Michael, Junio C Hamano. Thread: https://gitlist.dev/t/32340 ## David Michael, 2012-12-14 19:56 Subject: [PATCH 1/4] Support builds when sys/param.h is missing Message-ID: URL: https://gitlist.dev/e/CAEvUa7mHHy1%3DaOMfg4hNnDzod4zSnNHZN87isgf6Yxh5cRAf0Q%40mail.gmail.com ``` An option is added to the Makefile to skip the inclusion of sys/param.h. The only known platform with this condition thus far is the z/OS UNIX System Services environment. Signed-off-by: David Michael --- Makefile | 5 +++++ configure.ac | 6 ++++++ git-compat-util.h | 2 ++ 3 files changed, 13 insertions(+) diff --git a/Makefile b/Makefile index 736ecd4..8c3a0dd 100644 --- a/Makefile +++ b/Makefile @@ -165,6 +165,8 @@ all:: # Define NO_POLL if you do not have or don't want to use poll(). # This also implies NO_SYS_POLL_H. # +# Define NO_SYS_PARAM_H if you don't have sys/param.h. +# # Define NO_PTHREADS if you do not have or do not want to use Pthreads. # # Define NO_PREAD if you have a problem with pread() system call (e.g. @@ -1748,6 +1750,9 @@ endif ifdef NO_SYS_POLL_H BASIC_CFLAGS += -DNO_SYS_POLL_H endif +ifdef NO_SYS_PARAM_H + BASIC_CFLAGS += -DNO_SYS_PARAM_H +endif ifdef NO_INTTYPES_H BASIC_CFLAGS += -DNO_INTTYPES_H endif diff --git a/configure.ac b/configure.ac index ad215cc..317bfc6 100644 --- a/configure.ac +++ b/configure.ac @@ -699,6 +699,12 @@ AC_CHECK_HEADER([sys/poll.h], [NO_SYS_POLL_H=UnfortunatelyYes]) GIT_CONF_SUBST([NO_SYS_POLL_H]) # +# Define NO_SYS_PARAM_H if you don't have sys/param.h +AC_CHECK_HEADER([sys/param.h], +[NO_SYS_PARAM_H=], +[NO_SYS_PARAM_H=UnfortunatelyYes]) +GIT_CONF_SUBST([NO_SYS_PARAM_H]) +# # Define NO_INTTYPES_H if you don't have inttypes.h AC_CHECK_HEADER([inttypes.h], [NO_INTTYPES_H=], diff --git a/git-compat-util.h b/git-compat-util.h index 2e79b8a..ace1549 100644 --- a/git-compat-util.h +++ b/git-compat-util.h @@ -104,7 +104,9 @@ #endif #include #include +#ifndef NO_SYS_PARAM_H #include +#endif #include #include #include -- 1.7.11.7 ``` ## Junio C Hamano, 2012-12-14 23:41 Subject: Re: [PATCH 1/4] Support builds when sys/param.h is missing Message-ID: <7vpq2curpa.fsf@alter.siamese.dyndns.org> URL: https://gitlist.dev/e/7vpq2curpa.fsf%40alter.siamese.dyndns.org In-Reply-To: ``` David Michael writes: > An option is added to the Makefile to skip the inclusion of sys/param.h. > The only known platform with this condition thus far is the z/OS UNIX System > Services environment. > > Signed-off-by: David Michael > --- Hmm, makes me wonder if we can remove that inclusion everywhere instead. What definitions are we getting from it? 8695c8b (Add "show-files" command to show the list of managed (or non-managed) files., 2005-04-11) added it to show-files, and without the include the compilation failed due to lack of MAXPATHLEN, but you shouldn't be using that in the first place (we tend to either use PATH_MAX or lift the upper limit using strbuf these days.). 65bb491 (Add the ability to prefix something to the pathname to "checkout-cache.c", 2005-04-21) does the same to checkout-cache.c to grab MAXPATHLEN. bb233d6 (Add support for a "GIT_INDEX_FILE" environment variable., 2005-04-21) moves these two inclusions to cache.h; removing it bteaks compilation due to lack of MAXPATHLEN, ULONG_MAX, etc., I didn't check what else sys/parms.h was accicdentally slurping into the compilation, but ULONG_MAX is what we explicitly ask for by including either inttypes or stdint these days. Later 4050c0d (Clean up compatibility definitions., 2005-12-05) further consolidated the headers by moving inclusion around. This commit did not audit unused headers. I have this suspicion that nobody would notice if we simply stopped including the header. ``` ## David Michael, 2012-12-15 02:46 Subject: Re: [PATCH 1/4] Support builds when sys/param.h is missing Message-ID: URL: https://gitlist.dev/e/CAEvUa7kZE-p0oVV%2BJpXY4y-%3DieB-%3DTHTrO4hxzZsFYhLvgvbFw%40mail.gmail.com In-Reply-To: <7vpq2curpa.fsf@alter.siamese.dyndns.org> ``` Hi, On Fri, Dec 14, 2012 at 6:41 PM, Junio C Hamano wrote: > I have this suspicion that nobody would notice if we simply stopped > including the header. While I'm not aware of any subtleties it could be causing on other platforms, it does seem fine to drop sys/param.h on my test GNU/Linux systems. I can resend the series and just remove the include, if preferred. Thanks. David ``` ## Junio C Hamano, 2012-12-15 17:53 Subject: Re: [PATCH 1/4] Support builds when sys/param.h is missing Message-ID: <7va9tfurq1.fsf@alter.siamese.dyndns.org> URL: https://gitlist.dev/e/7va9tfurq1.fsf%40alter.siamese.dyndns.org In-Reply-To: ``` David Michael writes: > On Fri, Dec 14, 2012 at 6:41 PM, Junio C Hamano wrote: >> I have this suspicion that nobody would notice if we simply stopped >> including the header. > > While I'm not aware of any subtleties it could be causing on other > platforms, it does seem fine to drop sys/param.h on my test GNU/Linux > systems. > > I can resend the series and just remove the include, if preferred. I am sure the patch as posted is much safer, but I was hoping that folks on non-Linux platforms may say "I tried to compile with the include removed, and I get identical binaries from before". Until that happens, I would prefer queuing your patch as-is. Thanks. ```