On 27/08/07, Simetrical <Simetrical+wikilist(a)gmail.com> wrote:
Well, they're close enough to right about the
second point. How many
"patch, need-review" on Bugzilla? 149, over 50% more than the number
of shell bugs. And yet we have a couple of times as many developers
as sysadmins, at least.
There are various reasons for this, just as there are (in addition to
lack of shell users' time) reasons for the shell bugs being
unfulfilled, but I think it's important to note that "need-review"
means, "this bug has a patch which has not been reviewed by anyone
other than the patch author" - in other words, *anyone* with a
reasonable grasp of MediaWiki is free to leave comments on a possible
patch.
A patch may be for an area of the code base which nobody available to
review the patch is comfortable with changing, or able to test -
patches against texvc often lurk for quite some time, mainly because
if you don't have the environment set up, then compiling texvc is a
pain in the backside. In general, I don't like applying l10n patches
for right-to-left or Asian languages because I can pretty much
guarantee my text editor will cause it to go wrong, for example.
A patch may offer to solve a problem that doesn't need solving, or
that shouldn't be solved in the mainline code, or it may offer a
solution that hasn't really been thoroughly discussed. A patch may be
unacceptable in quality (bad code practice, bad documentation,
performance nightmare) or incomplete (schema change with no associated
updater provided). Some of the "patches" marked on BugZilla are not,
in fact, patches - comments with, "you need to change line X to line
Y" aren't actually very easy to review at all, for instance.
There are, no doubt, a multitude of counter-arguments to these points,
many of which are fair. I think the main reminder here is that if you
want your patch reviewed, *** keep pestering somebody *** - it may be
that your patch was noticed, but forgotten about.
Rob Church