>Has anyone thought to bundle up this discussion (before it goes off into
>meta-issue land) and forward it to Sam Leffler?  Presumably he had *some*
>reason for putting the stuff there in the first place, and on multiple
>platforms to boot.  My own observation was that it appeared that he was
>trying to keep most everything under one roof that was related, much as
>one might do with a package (e.g., I keep all my MH stuff under /usr/local/mh)
>.
>But since so much of the stuff had to do with spooling, I guess he decided
>that's where the main part of the package should go.  Admittedly I haven't
>asked him, and this may be an uninformed opinion on my part.  Caveat actor.

Here's an only slightly-less uninformed opinion, based on the e-mail and
work I've done with HylaFAX getting the alpanumeric pager support working
better.

The defaults for the HylaFAX distribution are designed for SGI systems, as
that's where Sam does all his development (big surprise), and he probably
never gave any though to how they fit into the BSD hierarchy.  So it's
probably not a surprise that the defaults aren't right for BSD systems.  My
guess is that he wanted to separate the server-side of the apps from the
client-side.

One other note - the configure process for HylaFAX has a separate entry for
the server backends (which happens to default to /var/spool/fax/bin), so it's
very easy to not have HylaFAX binaries under /var.  However, the spooling area
encompasses log files, queue files, _and_ config files.  So short of ripping
the distribution apart it's not possible to separate obvious spooling stuff
from the config files.  But since crontabs currently live in /var, there seems
to be some precedent for having some sorts of config files in /var.

--Ken