From: rst@ai.mit.edu (Robert S. Thau)
Date: Fri, 21 Jun 1996 18:43:13 -0400
Message-Id: <199606212243.SAA29406@volterra.ai.mit.edu>
To: new-httpd@hyperreal.com
Subject: Authoring support?
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

If you've got Netscape Gold or GNN press lying around, it might be fun
to try saving files to

   http://www.ai.mit.edu:8000/xperimental/sandbox 

I've tried it with both, and data from both seems to show up, but your
milage may vary.  It should let you in without authentication (there's
a No_auth line in a control file which allows this).

NB this is an experiment --- the server itself and the directory may
both vanish at any time.  

If you've got neither package, and you're really curious, I recommend
trying GNN Press first ---

  http://www.gnnhost.com/publish/upgrade/gnnpress.htm

GNN Press seems to be somewhat less flaky than at least the Linux beta
of Netscape Gold, which has a really annoying tendancy to hang on PUTs
if a prior PUT got any kind of an error.  Unfortunately, at least on
my screen, GNN Press's default fonts are way too small for comfort,
but that can be fixed, though it's a bit of a hassle.  (It also has a
tendancy to put up bogus "You may not be able to save a document from
this protected area" dialog boxes when you start to edit a document
that you fetched from the sandbox --- ignore them).

I'd like to recommend the W3C's authoring tool, Amaya, but it seems
they haven't released it yet.  Pity, that.

FWIW, the sandbox directory is back-ended by RCS, and uses the CGI
environment variables on each PUT for log entries (or tries to --- I
don't exactly trust it yet).  If you're brave enough to try saving an
image file, that should (there's that word again) get uuencoded before
being fed to RCS, so RCS doesn't choke.

(The engine here is a PUT back-end CGI script, which doesn't rely much
for support on the web server itself, except for a little paranoia
support which I tossed in along the lines I've been discussing, to
make it at least more difficult for randoms with CGI privilege on the
web server from faking up PUT requests into other peoples'
directories).

Hopefully, this code will be in distributable shape in the not too
distant future...

rst