Re: Bug tracking (was Re: +/- Inf for float8's)

Don Baccus wrote:
>
> How about ignoring whether or not a new solution is borne, or the SDM
> can be customized to fit your needs with less work than, and concentrating
> on what features you want to see?
>
> Has anyone thought about this, or was the last attempt such a failure that
> people have given it no further thought?
>

I don't know what the major code contributors need beyond the
TODO list. But I remember what went wrong with the older system
-- people would post non-bug issues, and in large numbers, as
bugs. And the system would "pend" those non-issues, assigning
them to core developers, who, at the time, were very busy
implementing MVCC and crushing real bugs by the hundreds. It
seems all Peter wants is a system whereby authorized users
(presumably those with CVS privileges) would have the ability to
post and close bugs. Perhaps such a system might have prevented
the duplicate work done recently on the "binary compatibility WRT
functional indexes" issue. Just from lurking, I think the core
developers' consensus was that anything which allows Joe User to
open tickets, without a "front-line" of advanced users/minor code
contributors willing to act as filters, would consume too much
time. People with great frequency ignore the note on the web-site
which reads "Note: You must post elsewhere first" with respect to
the pgsql-hackers list.

So I don't think it was an issue with the technology, but the
process. Although, from what I've read, I suspect ArsDigita is
light-years ahead of the Keystone software that was the
"PostgreSQL Bug Tracking System".

P.S.: I've been looking forward to seeing ArsDigita running on
postgresql.org for some time. I suspect there would be some
short-term pain, but substantial long-term gain. :-)

Mike Mascari