apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Stein <gst...@lyra.org>
Subject Re: cvs commit: apr/misc/unix errorcodes.c
Date Tue, 05 Jun 2001 17:30:15 GMT
Ease up, David. Cliff did the right thing here.

Yes, maybe the "no checking" isn't documented, but it has been policy for a
long while. Yes, we *should* document it. Nobody has got around to it is
all. I'm surprised that you aren't familiar with it.


On Tue, Jun 05, 2001 at 09:23:08AM +0100, David Reid wrote:
> >   * Remove the unnecessary parameter checks and the extra error codes that
> >     went along with them.  The APR policy is to segfault on a NULL
> parameter
> >     rather than silently returning some error code that the caller might
> >     not check anyway.
> Can't say I agree 100% with this, but if you say so, it must be.  BTW, where
> is this written/documented??
> >   * Also remove lots of unnecessary assertions (where the code would have
> >     segfaulted anyway, even without an explicit assert).  I've tried to be
> >     sure that every one I removed will result in a virtually immediate
> >     segfault anyway.  Ones that don't are the ones that are tricky to
> debug.
> >     If I've removed too many, say so and I'll put them back.
> Whatever...
> >   * Fix a misnamed APR_MEMORY_ASSERT -> APR_ASSERT_MEMORY, which was
> causing
> >     apr_assert_memory() never to be compiled.  Also fix a syntax error in
> that
> >     function that's been there since rev 1.1 of apr_sms.c, which no one's
> >     ever noticed because they never compiled it before.
> If you checked you'd see that the misnamed assert tag was added in revision
> 1.5 of the original file, and so the code that you claim was never built was
> building quite happily up till that point.  As for the screw up on the
> naming, don't believe I missed that!
> david

Greg Stein, http://www.lyra.org/

View raw message