> Thomas, we now get:
> select usesysid from pg_user union select null ;
> ERROR: type id lookup of 0 failed
> which not good either. Can you address this issue?
I'm almost back on-line (I hope) after massive hacker activity took out
the alumni pop server at Caltech. From looking through the hackers
mhonarc archive (hmm, don't much like that name "hackers" anymore; it's
cost us a _lot_ at work and taken me out of Postgres for more than a
month :( I see two issues for me to look at: the one above and the one
from Brett regarding a core dump from a mal-formed query. Will look at
both.
I have some additional patches in the parser area which continue the
type matching/coersion work from the last two months.
I've got patches to put 64-bit integers into the backend; they should
work for Alphas and at least some gcc-compiled 32-bit machines, but
we'll need beta testers to help get the configuration for our other
supported platforms. Once we have that for v6.4, we can also use this
type internally to implement additional types like numeric() and
decimal().
All mail sent to me from June 12 to now has been lost (I lost access to
the pop server and procmail went away, and so my .forward file reference
to procmail was broken and unfixable). It is not yet fixed, but should
be in a day or so. Talk to you then. In the meantime, I will be looking
at the mhonarc archives to keep up (great feature scrappy!)...
- Tom