OK, I guess I should contribute the solution I finally used. The
problem at hand was a 34 MB mailing list archive containing lots of
news excerpts (about 10000 messages). For an article I'm going to
write, I wanted to scan that archive for contributions concerning
certain topics.
My solution was this: First, using mutt, I stored that archive to a
maildir folder. I then created a sub-directory hashed/, and used a
small shell script (horribly inefficient) to move the message files
into a certain number of subdirectories under hashed/. That way,
file system overhead was under control again. I then indexed the
hashed/ subdirectory using glimpseindex, and finally used the
attached script to create hard links to the cur sub-directory, which
could then be read using mutt.
Note that this script has a kludge to emulate glimpse -l - that's
because glimpse -l dumps core on my system.
--
http://www.guug.de/~roessler/
#!/bin/sh --
glimpse -H . -1 -i "$@" | cut -f1 -d: | while read filename ; do
ln $filename cur/
done
