apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Klaus Keppler <klaus.kepp...@informatik.stud.uni-erlangen.de>
Subject Re: [PATCH] RESEND: fixing condvar bug under WIN32
Date Sat, 28 Aug 2004 21:38:53 GMT
> Unfortunately, given our current track record, posting patches to the
> bugdb isn't the best way to get them noticed, as I'm sure you have seen.

Hmm - I posted the bug in march to apr-dev - and was told to open
a bugzilla ticket. OK - I did so - but nothing happened. In the last
discussion about this specific bug it was said that the WIN32
implementation of mutexes and locking should be completely rewritten.
But nothing happened.
So - to prevent that things get lost - I wanted to send just some
kind of "hello? there's still a bug!" ;-)

> Whether or not to include a patch like this in APR 1.0.0 would of course
> be the prerogative of David Reid since he's the Release Manager.  But I
> can tell you that if I were RM, I'd say that a patch like this to fix a
> bug would just have to wait for APR 1.0.1.

For me it's no problem to always include a specially patched APR version
with my application. But I personally think that when the Apache Group
officially releases a "1.0" version of its APR to the world, and
some people get interested in APR and use it, they'll be disappointed
if those problems still exist (I needed nearly two working days to
find that the bug is not in my own application (as mostly...), but
in the underlaying APR).
The PR around APR will probably never be that big again like at
the 1.0 release.

> Build breakage might be one
> thing, but after no less than *six* release candidates, then something
> that's "just a bug" would seem to have little justification for holding
> up the release at this late date.

Yes, that's right - I should have reacted after RC1, but I just didn't
have the time to port my code from using apr-0.9.4 to apr-1.0.0rc[x].

Today I got this time (as you might have noticed on my amount of
apr-dev mailings ;-) - but I worked a whole day to get all things
working again with 1.0.00rc6 (included time for testing, seeing
the threading code crash and patching)

I won't worry if the patch won't be applied to 1.0.0, but at least in
a 1.0.1 release shortly after 1.0.0 it should be considered.

God - am I the only human on earth using conditional variables
with APR under WIN32? ;-)

Good night...

Klaus



Mime
View raw message