>
> > From: "Andrej Borsenkow" <[EMAIL PROTECTED]>
> > Date: Thu, 28 Jun 2001 13:58:35 +0400
> >
> > It checks only for specific option for IRIX and _FILE_OFFSET_BITS and
> > _LARGE_FILES macros. That is not enough; e.g. on our system:
> >
> > bor@itsrm2% getconf LFS_CFLAGS
> > -D_FILE_OFFSET_BITS=64 -D_LONGLONG -Kll64
>
> This wouldn't be Reliant UNIX, would it?
Yes. Wow!
(If it makes you jump
> through silly hoops like that just to access large files, no wonder
> the operating system is so rarely used. :-)
>
Be fair. Reasons why it is so rarely used are totally different. Actually,
the system is much better (IMHO of course) than some other commercial Unices
I had to work with, including Solaris and SCO. The above is bullet proof
method; actually, both _LONGLONG and -Kll64 are default.
> Anyway, can you suggest a simple test for this problem, and suggest
> what flags should be added for that OS? We could add a special case
> for it, just as there is already a special case for IRIX 6.2.
>
Assuming you are using native compiler, _FILE_OFFSET_BITS should be
enough. -Kll64 is not described anywhere; I suspect it is obsoltete version
of -Klonglong that is on by default. My question was mostly about getconf
usage.
> > why AC_SYS_LARGEFILE does not use getconf that is the
> > sdtandard interface to obtain {C,LD}FLAGS needed to compile
> program with LFS
> > support?
>
> I used that method originally, but it was a miserable failure in
> practice. For example, that method fails for Bash 2.04 on Reliant
> Unix with GCC; see
> <http://mail.gnu.org/pipermail/bug-bash/2001-March/001571.html>.
I know. I disabled LFS with GCC for zsh :-)
>
> And you can't easily cross-compile once you invoke getconf.
>
> I used to recommend `getconf LFS_CFLAGS', but I don't do it any more;
> in practice, it causes more trouble than it cures. Also, `getconf
> LFS_CFLAGS' will not be included in the next POSIX standard.
>
Ah, that is the argument. Is LFS supported by default then?
-andrej
