APR Status Report September '07

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

It seems that this report again coincides with new releases. Immediately
following our June report, apr 1.2.9/0.9.14, apr-util 1.2.8/0.9.13 and
apr-iconv 1.2.0 were all released smoothly as expected.

This month, APR 1.2.11/0.9.16, and APR-util 1.2.10/0.9.15 libraries were
released.  Both were re-rolled after the first try to pick up heavy activity
around the regression tests and to accommodate some of the platform tweaks.
Two interesting tweaks in the recent release may have interesting side
effects; APR determined to respect Mac OS/X's claim to handle only utf-8
filenaming conventions (as their UI, Aqua does) while on Win32 the process
creation was tweaked to mirror Unix's behavior for pipe inheritance and
therefore avoid leaking handles (in fact this leveraged the implementation
in OS2's APR code).  Apparently some third party libraries are negatively
impacted by this second adjustment, since they had relied on the side effects
of the old Win32-only behavior.  Work is in progress and a dialog is ongoing
to address those users concerns and provide them an alternate behavior which
is portable.

>From June to date in September, we had some 12 commits to apr-iconv/trunk,
64 commits to apr-util/trunk and 97 commits to apr/trunk (the future 1.3
or 2.0 version and first stop of all bug fixes on their way to a release
branch), by 14 active committers, many maintenance commits on the legacy
branches.  From early July through early August the bugs were brought down
to some 25 open incidents at a time, since then this has crept back up to
to the low 30's, below the previous average.  Some 36 bugs were closed,
and 6 tagged needing feedback, leaving 16 active bugs and 4 stale bugs
looking for some resolution.

Nothing much to report on community activity, two committers were added
in this period, davi and rpluem, both of whom were then active in this
period.  One request for commit karma was returned to the requester, with
a counter-request for that individual to submit patches.  Based on those,
they would be considered again for commit access at a later date.  This was
in contrast to an earlier policy, which brought in httpd, svn and similar
users with minimal scrutiny, if they came from any sibling community.  It
seems that policy may no longer fit the apr pmc's wishes, so I would expect
all commit karma moving forwards will follow the typical route of accepting
patches, granting committer status to frequent patch submitters, and then
considering active committers for PMC membership.

Respectfully submitted,
William A. Rowe, Jr.
APR Chairman