From: | Gregory Stark <stark(at)enterprisedb(dot)com> |
---|---|
To: | "Heikki Linnakangas" <heikki(at)enterprisedb(dot)com> |
Cc: | "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "PostgreSQL-development" <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Visibility map thoughts |
Date: | 2007-11-06 20:13:19 |
Message-ID: | 87mytr16ao.fsf@oxford.xeocode.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
"Heikki Linnakangas" <heikki(at)enterprisedb(dot)com> writes:
> I don't buy that. I believe at least on some architectures you'd get a
> word-long load+modify+store, and scribble the neighboring bytes.
Hm, I mis-remembered this bit of advice from the glibc info doc. I remembered
thinking it was strange when I read it but I guess my memory exaggerated how
strange it was:
.> In practice, you can assume that `int' is atomic. You can also assume
.> that pointer types are atomic; that is very convenient. Both of these
.> assumptions are true on all of the machines that the GNU C library supports
.> and on all POSIX systems we know of.
I suppose if we could keep count of tuples and a count of free space and use a
whole word. Map files would be 1M per 2G heap file (on an 8kb blocksize and
4-byte words). More complicated than necessary but I'm just thinking out loud.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
Ask me about EnterpriseDB's RemoteDBA services!
From | Date | Subject | |
---|---|---|---|
Next Message | Gregory Stark | 2007-11-06 20:51:59 | Re: Weird type selection choice |
Previous Message | Gregory Stark | 2007-11-06 20:04:29 | Re: Visibility map thoughts |