Re: svn commit: r348499 - in /apr/apr/trunk: CHANGES poll/unix/select.c test/testpoll.c

[Joe Orton <jo...@redhat.com>]

On Wed, Nov 23, 2005 at 05:38:58PM -0000, Brian Pane wrote:
> Author: brianp
> Date: Wed Nov 23 09:38:47 2005
> New Revision: 348499
> 
> URL: http://svn.apache.org/viewcvs?rev=348499&view=rev
> Log:
> Bugfix for apr_pollset_poll() on systems that implement pollsets
> using select(2): properly compute the number of signalled desciptors
> when one or more of them are both readable and writable.
> Submitted by: Dror Shilo <Dr...@ericom.com>, Gerry <ge...@everythingsucks.co.uk>
> Reviewed by: Brian Pane

I think the select()-based apr_poll() implementation has the same bug.

joe

Re: svn commit: r348499 - in /apr/apr/trunk: CHANGES poll/unix/select.c test/testpoll.c

[Justin Erenkrantz <ju...@erenkrantz.com>]

On Fri, Nov 25, 2005 at 08:02:10PM -0800, Greg Stein wrote:
> On Fri, Nov 25, 2005 at 07:55:28PM -0800, Brian Pane wrote:
> >...
> > A procedural question on backports: are 0.9/1.0/1.1/1.2 using C-T-R
> > rules, or R-T-C?
> 
> IMO, it should always be CTR.

Yup.  APR is governed by its backwards-compatibility guidelines.  If a
backported commit breaks that, it'll have to be reverted.  Therefore,
there isn't a strong need for RTC for APR.  -- justin

Re: svn commit: r348499 - in /apr/apr/trunk: CHANGES poll/unix/select.c test/testpoll.c

[Greg Stein <gs...@lyra.org>]

On Fri, Nov 25, 2005 at 07:55:28PM -0800, Brian Pane wrote:
>...
> A procedural question on backports: are 0.9/1.0/1.1/1.2 using C-T-R
> rules, or R-T-C?

IMO, it should always be CTR.

If somebody disputes a fix for an older release, then put the burden
on them to do so. I much prefer more freedom in applying fixes. Put
the work on keeping stuff *out* than on bringing stuff *in*.

Cheers,
-g

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

Re: svn commit: r348499 - in /apr/apr/trunk: CHANGES poll/unix/select.c test/testpoll.c

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

Brian Pane wrote:
> 
> Confirmed, and I've just committed a fix.
> 
> A procedural question on backports: are 0.9/1.0/1.1/1.2 using C-T-R
> rules, or R-T-C?

Don't waste your time on 1.0/1.1, they were DOA with the arrival of 1.2.

Remember our back-compat rules, there's no reason for someone to use 1.0
when 1.2 is out, programs built for 1.0 should be binary/behavior compat
to any 1.x release.

AFA 0.9 - it's a moving target (minimal versioning enforcement), so feel free
to C-T-R there, too.

Bill

Re: svn commit: r348499 - in /apr/apr/trunk: CHANGES poll/unix/select.c test/testpoll.c

[Brian Pane <br...@apache.org>]

On Nov 25, 2005, at 5:38 AM, Joe Orton wrote:

> On Wed, Nov 23, 2005 at 05:38:58PM -0000, Brian Pane wrote:
>> Author: brianp
>> Date: Wed Nov 23 09:38:47 2005
>> New Revision: 348499
>>
>> URL: http://svn.apache.org/viewcvs?rev=348499&view=rev
>> Log:
>> Bugfix for apr_pollset_poll() on systems that implement pollsets
>> using select(2): properly compute the number of signalled desciptors
>> when one or more of them are both readable and writable.
>> Submitted by: Dror Shilo <Dr...@ericom.com>, Gerry  
>> <ge...@everythingsucks.co.uk>
>> Reviewed by: Brian Pane
>
> I think the select()-based apr_poll() implementation has the same bug.

Confirmed, and I've just committed a fix.

A procedural question on backports: are 0.9/1.0/1.1/1.2 using C-T-R
rules, or R-T-C?

Thanks,
Brian