On Wednesday, August 2 at 04:57 PM, quoth Thomas Roessler:
Seeing the repeated "delete header cache" problems, I'm wondering if it would be worth implementing an "remove existing cache" command line option, or maybe even an interactive function. What do people think?
Mmmm... well, it's a stab, but I don't think it would solve the number of people complaining about such things. Rather than responding "it's crashing because the cache is corrupted, run `rm -f $cache`" we'd be responding "it's crashing because the cache is corrupted, run `mutt --delete-cache`", so it doesn't really *solve* much other than automatically figuring out where the cache is stored.
What I think would be really useful is if mutt did two things:1. If the cache is the wrong version or is corrupted or whatever: DELETE IT automatically! There's no reason that the user has to watch mutt die (or better yet, wonder why things are so slow, when mutt is silently ignoring the cache as unusable rather than delete such precious unusable temporary data) when the problem is *temporary* data.
2. Stop storing pointers in the cache, and/or stop pulling them out and dereferencing them. The way mutt currently does it is extremely fragile and prone to exactly these sorts of "it crashed!" issues. There's no reason mutt *should* crash, so it really should stop it.
~Kyle --It is a dogma of faith that the demons can produce wind, storms, and rain of fire from heaven.
-- St. Thomas Aquinas
Attachment:
pgpiAyTAEjI6W.pgp
Description: PGP signature