[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: FYI: removing HAVE_FCNTL_H tests
From: |
Jim Meyering |
Subject: |
Re: FYI: removing HAVE_FCNTL_H tests |
Date: |
Thu, 14 Jul 2005 14:45:51 +0200 |
Bruno Haible <address@hidden> wrote:
> Jim Meyering wrote:
>> > I've just removed all tests for HAVE_FCNTL_H from coreutils.
>> > It's been gone in at least one place since coreutils-5.0 (2004-04-02)
[actually, that should have been 2003-04-02, not 2004]
>> The same applies to HAVE_UNISTD_H and unistd.h
>
> One platform still does not have <unistd.h>: It's Woe32 with the MSVC
> compiler. It's unfortunately a major platform, and still alive.
Perhaps it's not a popular enough build environment for packages using
gnulib. Several existing modules include <unistd.h> unconditionally:
lib/argp-parse.c
lib/chdir-long.c
lib/fts.c
lib/getpass.c
lib/openat.c
lib/poll.c
lib/utime.c
and yet I haven't heard any complaints.
> Also, <unistd.h> is less dependable for GNU programs than <fcntl.h>,
> because someone creates a brand new operating system different in style
> from Unix, he will typically implement the ISO C headers but not POSIX
> <unistd.h>. (Some recent operating systems like BeOS or MacOS X have
> a Unix underneath, but others don't.)
IMHO, a new OS that does not provide a POSIX header like <unistd.h>
does not deserve our consideration -- and isn't likely to get much
from any other development community, either.
> So for the sake of ease of porting to new OSes, I would leave HAVE_UNISTD_H
> in place.
Thanks for the info.
I've made a note to wait a year.
If there have still been no complaints,
I'll remove them then.