CommitFest Status Summary - 2010-02-03

Here's an overview of where we stand with the remaining 14 patches,
according to my best understanding of the situation.

* Fix large object support in pg_dump - I haven't looked at this, but
it seems like it's relatively close to being ready for commit. We
need to get this one done as it is a release blocker (IMO).
* Remove gcc dependency in definition of "inline" functions - We have
been waiting for an updated patch since January 19th. If there is no
update in the next few days, I will marked this as Returned with
Feedback.
* Faster CREATE DATABASE by delaying fsync - There seems to be a
consensus that this is doing something sensible, but we're still
arguing about how things should be named and documented. Doesn't seem
like anything insurmountable, though.
* More frame options in window functions - Tom has listed himself as
the committer on this one, so I infer that he is planning to commit it
after suitable editorialization. This is a big patch.
* Writeable CTEs - Another big patch. As per today's discussion, at a
minimum, it still needs some cleanup work and quite a bit of
additional documentation. I have set it back to Waiting on Author.
Even once that's fixed, I suspect this is going to need some attention
from Tom prior to commit.
* Provide rowcount for utility SELECTs - There's some debate about
whether this is a good idea, but I believe most people are in favor of
it, provided that we can cover all cases. This one really needs some
more review.
* Make PostgreSQL binaries use the new PQconnectdbParams() libpq
functions - I believe Joe Conway is working on this.
* rbtree - I have done a lot of work reviewing this, and Mark
Cave-Ayland has done some work on it, too. But there are some
unanswered performance questions that need to be addressed before
commit. This is another one that could really use some more eyes on
it.
* knngist - The third remaining big patch. Mark Cave-Ayland
volunteered to review this one, too, but so far no review has been
posted on -hackers. I know that the PostGIS folks would really like
to have this, but time is growing short.
* plpython3 - It doesn't seem likely that this will go into core for
9.0, but Nathan Boley is going to review it anyhow, which is a good
thing.
* Add on_trusted_init and on_untrusted_init to plperl - Still arguing
about the GUC names and details, but seems to be more or less on
track.
* Package namespace and Safe init cleanup for plperl - From the
discussion, this one seems to be in pretty good shape, too.
* Listen / Notify rewrite - The fourth and final major patch remaining
for this CommitFest. There are still ongoing discussions about the
concurrency handling in this patch, and some other open issues, too,
like limiting the payload to ASCII. I know this is another big
feature people would like to see in, but we're running out of time to
get it stabilized.
* xpath non-nodeset result enabling - We've been waiting on an updated
patch since January 17th. On the 28th, Scott Bailey said he might
take a look at it, but we've heard nothing since. If there is no
update in the next couple of days, I will mark this Returned with
Feedback.

...Robert