Re: [DOCS] Re: [HACKERS] vacuum problem

From: Bruce Momjian <maillist(at)candle(dot)pha(dot)pa(dot)us>
To: lockhart(at)alumni(dot)caltech(dot)edu (Thomas G(dot) Lockhart)
Cc: tih(at)nhh(dot)no, pgsql-hackers(at)postgreSQL(dot)org, docs(at)postgreSQL(dot)org
Subject: Re: [DOCS] Re: [HACKERS] vacuum problem
Date: 1998-08-28 14:25:17
Message-ID: 199808281425.KAA05957@candle.pha.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> > By the way, PostgreSQL somehow seems to have become significantly
> > faster for my use sometime over the last month or two. For the select
> > and update queries I regularly execute, which generally involve two
> > or three tables and ditto indices, I'm seeing what feels like twice
> > the speed of what I got before -- and I've been increasing the amount
> > of data in my tables without any schema changes or index additions!
>
> Without knowing the real reason, I'm going to jump in and have the type
> coersion code take credit for this *grin*.
>
> In particular, it _may_ do a better job of matching up indices with
> queries.

So, you want to take credit for it. :-)

You can. There is nothing I did to speed things up except to remove man
sequential scans of system tables from the code. I can't imagine that
causing the speedups he is reporting.

>
> Are there other reasons why things may have gotten faster?

It would be interesting to see if the old code did not use indexes, and
the new stuff does, and if type conversion was needed in those queries.

>
> It's about the time to start working on release notes (Bruce?), and
> perhaps this could be quantified and mentioned...

Yes, this weekend, perhaps tomorrow. I know people are waiting. Once I
do it, I have to add every additional change, and that is a pain, so I
wait until near the end.

Can you work on the regression tests? My oidname,... removal is a
problem.

>
> btw, the release notes are in sgml (doc/src/sgml/release.sgml) and all
> previous notes and detailed change lists I could find have been put into
> there. The notes for the next release can look very similar to what is
> there already, and there is already a section set aside for it.

OK, I see it. Looks like I am going to have to learn sgml. I have no
way of viewing it, so I will have to do my best, and you can tell me how
it looks. Is there an sgml->something conversion tool I can use for
testing?

--
Bruce Momjian | 830 Blythe Avenue
maillist(at)candle(dot)pha(dot)pa(dot)us | Drexel Hill, Pennsylvania 19026
+ If your life is a hard drive, | (610) 353-9879(w)
+ Christ can be your backup. | (610) 853-3000(h)

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Thomas G. Lockhart 1998-08-28 14:28:27 Re: [HACKERS] long long int and printf format for int8.c
Previous Message Bruce Momjian 1998-08-28 14:19:50 Re: [HACKERS] regression failures]