This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH][BZ #14292] Do not inverse seek in fmemopen_seek.
- From: OndÅej BÃlka <neleai at seznam dot cz>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Wed, 29 May 2013 09:27:38 +0200
- Subject: Re: [PATCH][BZ #14292] Do not inverse seek in fmemopen_seek.
- References: <20130520154411 dot GA24852 at domone dot kolej dot mff dot cuni dot cz> <20130528230147 dot 58AC82C05D at topped-with-meat dot com>
On Tue, May 28, 2013 at 04:01:47PM -0700, Roland McGrath wrote:
> It sure looks like a straightforward bug. Unfortunately it's entirely
> conceivable that some program has come to depend on the bogus behavior.
> The conservative approach to fixing it would be to define a new symbol
> version for fmemopen, so the old one behaves as it did before. I'm a bit
> ambivalent about whether we should bother with that.
>
> Some code archeology shows that the libio implementation of fmemopen
> (i.e. the only one that's ever been used on GNU/Linux) always behaved this
> way (though the original fmemopen implementation from when I invented it in
> the old GNU stdio never did). So I'm inclined to say we do indeed need to
> keep the compatibility version just in case.
>
As I am fixing several bugs concerning fmemopen I will post separate
patch for versioning that should be applied first