<<< Date Index >>>     <<< Thread Index >>>

Re: header cache (was Re: Why isn't it part of mutt "proper")



2004-08-19T16:16:13 Thomas Glanzmann:
> rm /tmp/sithglan-hcache

__OH__ --- sound of me smacking my head, hard, a meaty thud. Not to
worry, I wasn't using it anyway.

Of course.

I had firmly lodged in my brain that I'd never, ever, under any
circumstances have to wait for ages to open a humongous maildir ever
again. What a silly bunt I am:-).

Seriously, though, I think what you have implemented right now is
precisely right, it works well, it's robustly correct, and the
right-feeling fix, should the cache get sufficiently larded up with
stale stuff, is to delete the cache, rather than write a cleaner.
The cleaner wouldn't be widely used, so over time it'd be a
maintenance problem --- it has to have an intimately correct
knowlege of the precise database format, which I've just realized is
not _quite_ as trivial as I'd at first assumed:-).

> But, if you like to, I could of course write a maildir-header
> cache wiper, too.

No, unless there's some fiendishly clever way of somehow integrating
it right into mutt and ensuring that everybody who uses the cache at
all runs the cleaner, let's not go there at all; it'll be a
recurring problem when it falls out of perfect alignment with the
real database interaction code.

Thanks once again,

-Bennett

Attachment: pgpkxLR8LpA9c.pgp
Description: PGP signature