> Here is the patch that replace pg_autovaccum catalog with reloptions. I
> refactored the reloptions.c to support multiple parameters and made the
> action of adding a new option an easy task. I'm testing it yet, so don't
> expect it to work properly. I'll prepare docs as soon as I finish the
> tests. Do i have to prepare some regression tests?
> I don't provide a pg_autovacuum view as suggested by Itagari-san  but
> if others agree that we need it, I will work on it. I don't if we need a
> function (wrapper around getRelOption()) to get an option from
> reloptions array.
> I add an ugly-hack to \d+ foo. IMHO, it'll be good to know what options
> are used by table/index foo (we already do it for oids) but I'm not
> happy with my suggestion.
> I move RelationGet*() functions from rel.h. That's because we need some
> knowledge that's only in reloptions.c (getRelOptions). But I want to
> avoid including reloptions.h at some files.
> PS> don't forget to remove include/catalog/pg_autovacuum.h
Several things related to this patch have been committed:
(and several follow-on commits)
What still remains to be done for 8.4?
In response to
pgsql-hackers by date
|Next:||From: Simon Riggs||Date: 2009-01-11 03:03:24|
|Subject: Re: Hot standby, slot ids and stuff|
|Previous:||From: Bruce Momjian||Date: 2009-01-11 02:09:58|
|Subject: Documenting pglesslog|