regular expressions are powerful, but even so it would be a
pain in the rear end to make your ordinary patterns ignore
the possible presence of html comments. what's even worse is
that eudora's r.e. engine is buggy, so not all of the power
is really available.
let's think bigger here. today, the problem is html comments.
tomorrow it may be something else. what would be better than
a feature to remove html comments, is a more general feature
allowing other pre-processing to be specified.
suppose you could apply a series of regular expression substitutions,
before doing the filtering?
suppose you could apply different pre-processing depending on
which filters matched, and then re-apply other filters?
why not create a scripting language for filtering messages?
what you'd end up with would probably be a perl-like language.