Re: What's happening with pgsql-committers?
От | Alex Pilosov |
---|---|
Тема | Re: What's happening with pgsql-committers? |
Дата | |
Msg-id | Pine.BSO.4.10.10010072133460.19791-100000@spider.pilosoft.com обсуждение исходный текст |
Ответ на | Re: What's happening with pgsql-committers? (The Hermit Hacker <scrappy@hub.org>) |
Список | pgsql-hackers |
On Sat, 7 Oct 2000, The Hermit Hacker wrote: > On Tue, 3 Oct 2000, Tom Lane wrote: > > > Peter Eisentraut <peter_e@gmx.net> writes: > > > Has anybody been getting pgsql-committers messages the last few days? > > > > Coming through fine for me (at least when hub.org isn't wedged > > completely, which it was several times over the weekend...) > > We hit a corruption of -general again this past week (I've really gotta > dive into getting the postgresql backend working for majordomo instead of > Berkeley DB :( ) ... it was causing a perl process of 900+Meg to run, > which caused the load avg to climb unreasonably :( Oooo, ooo, how about instead writing a ndbm/gdbm/whateverdb wrapper for postgres? Configurable by some file in /etc, where it would establish correspondence between "file" argument to dbopen() and a postgres view. This would rock for converting applications which only support dbm mappings (such as sendmail) to use postgres transparently. The only problem I see with it is db->fd() function which returns a filedescriptor for the file, which some apps rely on to do flock/fcntl. On other hand, locking of that kind will be unnecessary for postgres, so it can return fd of some file in /tmp... There's an attempt to do this at http://www.is.kiruna.se/~goran/ldap/arkiv/ but its very raw and not configurable. -alex
В списке pgsql-hackers по дате отправления: