Natterings about history files
davidsen at tmr.com
Fri Mar 2 04:42:55 UTC 2001
In article <ylhf1e5vv0.fsf at windlord.stanford.edu>,
Russ Allbery <rra at stanford.edu> wrote:
| bill davidsen <davidsen at tmr.com> writes:
| > Perhaps this could be delayed until some time after the article (or
| > cancel for the article) is received, to avoid lookup delays. But after a
| > day or so, it's unlikely that the info will be needed often enough to
| > make the performance hit offset the gain in instant expire.
| I went down that road too in my head, and then realized that the problem
| is that if you're accepting a new article that you've never seen before,
| you have to do a lookup in *all* your history files, or otherwise you may
| as well not keep them. So it's not true that older history files get
| fewer lookups, unfortunately; new articles are the common case of stuff
| that makes it through the cache. :/
This is true, and for transit machines it depends on the cost of the
multiple lookups, purely. I believe that there is a flavor of perfect
hash which gives you either a hit with value or a miss, but I can't find
the info, so perhaps I'm wishing rather than remembering. But the extra
files I was envisioning were only for articles no longer on the server,
and don't change, so they lend themselves to a perfect hash if a perfect
hash suits our needs.
I don't know if this is a mirage or a solution I don't see clearly, so
feel free to ignore it. I would feel bad not to mention it, since it may
trigger someone else to a useful idea. And having said it, unless I
think of something pro or con I'm going to drop it for now.
bill davidsen <davidsen at tmr.com>
CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
More information about the inn-workers