Re: add more frame types in window functions (ROWS)

2009/12/7 Greg Smith <greg(at)2ndquadrant(dot)com>:
> Which means that views created in the window test could absolutely cause the
> rules test to fail given a bad race condition.  Either rules or window needs
> to be moved to another section of the test schedule.  (I guess you could cut
> down the scope of "rules" to avoid this particular problem, but hacking
> other people's regression tests to work around issues caused by yours is
> never good practice).  I also agree with Andrew's sentiment that including a
> view on top of the new window implementations is good practice, just for
> general robustness.

I've agreed window and rules cannot be in the same group if window has
view test.

> It looks like a lot of progress has been made on this patch through its
> review.  But there's enough open issues still that I think it could use a
> bit more time to mature before we try to get it committed--the fact that
> it's been getting bounced around for weeks now and the regression tests
> aren't even completely settled down yet is telling.  The feature seems
> complete, useful, and functionally solid, but still in need of some general
> cleanup and re-testing afterwards.  I'm going to mark this one "Returned
> with Feedback" for now.  Please continue to work on knocking all these
> issues out, this should be a lot easier to get committed in our next CF.

OK, Andrew, thank you for collaborating on this. This fest made my
patch to progress greatly. More comments from others on the memory
leakage issue are welcome yet.

Regards,

--
Hitoshi Harada