Re: mailing list archiver chewing patches
Matteo Beccati <php@beccati.com>
From: Matteo Beccati <php@beccati.com>
To: Dimitri Fontaine <dfontaine@hi-media.com>
Cc: David Fetter <david@fetter.org>, Aidan Van Dyk <aidan@highrise.ca>, Magnus Hagander <magnus@hagander.net>, Dave Page <dpage@pgadmin.org>, Abhijit Menon-Sen <ams@toroid.org>, Alvaro Herrera <alvherre@commandprompt.com>, Andrew Dunstan <andrew@dunslane.net>, Tim Bunce <Tim.Bunce@pobox.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2010-01-16T13:21:58Z
Lists: pgsql-hackers
Il 16/01/2010 11:48, Dimitri Fontaine ha scritto: > Matteo Beccati<php@beccati.com> writes: >> Anyway, I've made further changes and I would say that at this point the PoC >> is feature complete. There surely are still some rough edges and a few >> things to clean up, but I'd like to get your feedback once again: >> >> http://archives.beccati.org > > I've been clicking around and like the speedy feeling and the Thread > index appearing under any mail. Also getting the attachments seems to be > just working™. I've also checked than this "local thread" works on month > boundaries, so that you're POC is in a way already better than the > current archives solution. Thanks for the feedback. > Only missing is the search, but we have tsearch and pg_trgm masters not > far away… I haven't even looked at it as I was under the impression that the old engine could still be used. If not, adding search support should be fairly easy. >> You will find that pgsql-general and -hackers are subscribed and getting >> messages live, wihle -hackers-history and -www have been imported from the >> archives (about 200k and 1.5k messages respectively at 50 messages/s). > > Tried clicking over there and very far in the past indexes show no > messages. Here's an example: > > http://archives.beccati.org/pgsql-hackers-history/1996-09/by/thread Yeah, there are a few messages in the archives with a wrong date header. The list is generated using from min(date) to now(), so there are holes. At some point I'll run a few queries to fix that. >> Also, I'd need some help with the CTE query that was picking a wrong plan >> and led me to forcibly disable merge joins inside the application when >> executing it. Plans are attached. > > Sorry, not from me, still a CTE noob. Actually the live db doesn't suffer from that problem anymore, but I've able to reproduce the issue with a few days old backup running on a test 8.5alpha3 instance that still has a stock postgresql.conf. Cheers -- Matteo Beccati Development & Consulting - http://www.beccati.com/