Re: Pluggable Indexes (was Re: rmgr hooks (v2))

On Thu, Jan 22, 2009 at 9:15 AM, Heikki Linnakangas
<heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
>> Immediate use cases for me would be
>>
>> * ability to filter WAL records based on database or relation
>
> This patch isn't enough to allow the catalog lookups. Without the catalog
> lookups, you might as well implement that as an external tool, like
> pglesslog.

The fact the patch does not do anything that anyone might ever want is
not a sufficient grounds for rejecting it. If it were, zero patches
would ever get accepted. If additional changes are needed, Simon or
someone else can send a patch later with those changes.

Much ink has been spilled in this space over the size and difficulty
of reviewing Simon's hot standby patch, on the grounds that it is big
and changed many things. Of course, Simon did submit an earlier
version of this patch that was less big and changed fewer things, and
it was never committed even though Simon responded to all of the
review comments. In fact, even after you took the time to split it
back out again, and even after acknowledging that the split-out part
was good code and independently useful, you never committed THAT
either. And so here we sit in limbo.

If you now reject this patch because it is small and changes too few
things, then will you reject his next patch that is more comprehensive
on the grounds that the patch is now too big to review?

I wonder what Simon has to do to get a patch committed. It's been
four months since he started submitting patches, and so far the only
thing that's been committed is the pg_stop_backup() wait bug fix. If
the code were bad or required a lot of fixing to get it in committable
form, that would be completely understandable but no one is alleging
that.

...Robert