On Thu, Feb 24, 2022 at 4:18 PM <wrowe@apache.org> wrote:
>
> Author: wrowe
> Date: Thu Feb 24 22:18:42 2022
> New Revision: 1898399
>
> URL: http://svn.apache.org/viewvc?rev=1898399&view=rev
> Log:
> Replace PCRE with PCRE2 where it is available
>
> *) Support PCRE2 with buffers reuse through Thread Local Storage.
> trunk patches:
> 2.4.x patches:
> https://github.com/apache/httpd/pull/289.diff
> (PR: https://github.com/apache/httpd/pull/289)
> - +1: ylavic, rpluem, covener, steffenal
> + +1: ylavic, rpluem, covener, steffenal, wrowe
> ylavic: This backport proposal inludes the PCRE2 backport already accepted,
> we can apply this one instead or the original one first and then
> this one (I'd have to rebase the github PR first).
All my testing this week was against PR289 rollup, so it should just
be a matter of doing a merge origin/trunk. I did want to keep the
svn:mergeinfo sync'ed, so I'd merged using the individual commits
after comparing that to trunk.
I noticed we test for PCRE_DUPNAME which is rather silly now,
since ./configure rejects it, and anything even approaching rev 6.6
(the last version that didn't include it) would be incredibly unwise.
We don't use any HAVE_PCRE_DUPNAME macro, so that whole
test could clearly be moved back up to the revision test, it probably
makes sense to reject anything below the 8.x series in the first place.
>
> Author: wrowe
> Date: Thu Feb 24 22:18:42 2022
> New Revision: 1898399
>
> URL: http://svn.apache.org/viewvc?rev=1898399&view=rev
> Log:
> Replace PCRE with PCRE2 where it is available
>
> *) Support PCRE2 with buffers reuse through Thread Local Storage.
> trunk patches:
> 2.4.x patches:
> https://github.com/apache/httpd/pull/289.diff
> (PR: https://github.com/apache/httpd/pull/289)
> - +1: ylavic, rpluem, covener, steffenal
> + +1: ylavic, rpluem, covener, steffenal, wrowe
> ylavic: This backport proposal inludes the PCRE2 backport already accepted,
> we can apply this one instead or the original one first and then
> this one (I'd have to rebase the github PR first).
All my testing this week was against PR289 rollup, so it should just
be a matter of doing a merge origin/trunk. I did want to keep the
svn:mergeinfo sync'ed, so I'd merged using the individual commits
after comparing that to trunk.
I noticed we test for PCRE_DUPNAME which is rather silly now,
since ./configure rejects it, and anything even approaching rev 6.6
(the last version that didn't include it) would be incredibly unwise.
We don't use any HAVE_PCRE_DUPNAME macro, so that whole
test could clearly be moved back up to the revision test, it probably
makes sense to reject anything below the 8.x series in the first place.