I just poked at this a bit myself to see how the patch looked. There's
just over 4000 lines in the diff. Even though 1/4 of that is tests,
which is itself encouraging, that's still a good sized feature. The
rate at which code here has still been changing regularly here has me
nervous about considering this a commit candidate right now though. It
seems like it still needs a bit more time to have problems squeezed out
still.
Two ideas I was thinking about here:
-If you take a step back and look at where the problem parts of the code
have been recently, are there any new tests or assertions you might add
to try and detect problems like that in the future? I haven't been
following this closely enough to have any suggestions where, and there
is a lot of error checking aimed at logging already; maybe there's
nothing new to chase there.
-Can we find some larger functions you haven't tested this against yet
to throw at it? It seems able to consume all the cases you've
constructed for it; it would be nice to find some brand new ones it's
never seen before to check.
This has made a lot of progress and seems it will be a good commit
candidate for the next CF. I think it justs a bit more time than we
have left in this CommitFest for it right now, particularly given the
size of the patch. I'm turning this one into "returned with feedback",
but as a mediocre pl/pgsql author I'm hoping to see more updates still.
--
Greg Smith 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us
