Proposal for a MIME mapping spec

[Mike Hearn]

On Wed, 2004-07-07 at 16:43 -0700, Keith Packard wrote:
> Well, I wouldn't do it that way, but I suppose it's possible.  Instead, I
> suggest that applications which read the database first check timestamps to
> see if the cached version is out of date and regenerate the cache at that
> point themselves.  A few stats at program startup time to avoid some bug
> reports and mildly confused users seems like a good trade to me.

I see what you're getting at, and you're probably right - sometimes the
system will go wrong and users will get confused, but assuming we have
enough checkpoints (at desktop login etc) where the cache can be
regenerated I think that can be mitigated.

The problem with having apps do it themselves is that compiling this
cache could easily take several seconds, maybe much longer if you have
lots and lots of programs involved and a slow machine. So then you want
GUI feedback, but you want that to use the same toolkit as the rest of
the app and so it rapidly gets a lot more complicated than just running
it once at install time.

It also requires the cache file to be world writeable, or to be using
suid root binaries, or to move the cache file to somewhere less obvious
where admins may not see it, or to depend on SELinux so every app can
regenerate it.

thanks -mike