Daniel Farina <daniel(at)heroku(dot)com> writes:
> Thinking unnecessary. Motion is progress. Here is a patch that uses
> this exact plan: pgport for the tables, broken out into a header file
> that is included in the building of libpgport. I have confirmed by
> objdump -t that multiple copies of the table are not included in the
> postgres binary and the bloat has not occurred.
Applied with minor adjustments; mainly, I cleaned up some additional
traces of the old way of building pg_resetxlog and pg_controldata.
regards, tom lane
In response to
pgsql-hackers by date
|Next:||From: Josh Berkus||Date: 2012-02-29 01:26:48|
|Subject: Re: 3rd Cluster Hackers Summit, May 15th in Ottawa|
|Previous:||From: Tom Lane||Date: 2012-02-28 22:35:14|
|Subject: Parameterized-path cost comparisons need some work|