Re: libiberty requirements and ISO C90

[Ian Lance Taylor <ian at airs dot com>]

Mark Kettenis <mark.kettenis@xs4all.nl> writes:

>    From: Ian Lance Taylor <ian@airs.com>
>    Date: 15 May 2005 23:20:14 -0400
> 
> 
>    Well, we require an ISO C90 compiler; do we require ISO C90 libraries?
>    If we require the libraries, then we can remove a number of files from
>    libiberty, at least atexit.c, memchr.c, memcmp.c, memcpy.c, memmove.c,
>    memset.c, snprintf.c, strchr.c, strerror.c, strncmp.c, strrchr.c,
>    strtol.c, strtoul.c, vfprintf.c, vsprintf.c.  If we don't require the
>    libraries, then we can't assume that <string.h> declares strerror.  In
>    fact, technically we can't even assume that <string.h> exists,
>    although we do currently have a few cases where it is included without
>    being protected by #ifdef HAVE_STRING_H.
> 
> Hmm, it seems that snprintf() wasn't in ISO C90 since it's mentioned
> as a new addition in ISO C99 according to google.  At least HP-UX
> 10.20 doesn't have it.  And GDB still builds on vax-dec-ultrix4.0 for
> which I'm not sure that it has a full ISO C90 library.  So I hope
> you're not seriously proposing to remove these functions from
> libiberty.

I'm not.  And, you're right, snprintf shouldn't be in the list anyhow.

Ian