Re: Pike 8.0: Sql.pgsql objects don't get garbage collected?

Jeff Hungerford wrote:
>It looks like that fixed it!  I've been running a number of different
>tests, and I haven't seen any errors

>I don't seen any connection leaks, and everything looks good.
>Thanks for fixing those issues!

Well, thank *you* for running the stresstests.
Doing comprehensive stresstests/unittests for database(drivers) is hard,
especially if threading/races are involved.  I'd love to have added unittests
to the Pike repository, but alas, without an available database, it's of no
use to write those.

> (except for seeing
>"Attempting to clone an unfinished program" on startup from time to time).

It should be possible to write a synthesised self-contained unittest for this
that has a high probability of triggering the race conflict (not using the
pgsql driver).  And no, I'm not volunteering to do that at the moment.
-- 
Stephen.