Improved mail archive prototype

[Owen Taylor <otaylor redhat com>]

[ 
  I sent the following to the gnome-hackers mailing list,
  and didn't get much response. (I guess everybody there is
  already subscribed to every mailing list.) So, I'm trying
  again in a wider forum.
]

I spent a bit of time today setting up some better archiving
on mail.gnome.org.

A prototype of the result for a subset of the lists is at:

 http://mail.gnome.org/archives/

This is using MHonArc instead of the built-in pipermail archiver,
without some loving to make it look better than the typical MHonArc
"what the dog brought in".

Improvements include:

 - searching (!) - Uses the namazu search engine, which is
   cool because it understands mhonarc formatted .html files
   and can do field searches, e.g.:

   +subject:gtkclist

   It also didn't require setting up a database backend, and
   worked out of the box, which made me happy.

 - Proper mime handling (see, e.g. 

   http://mail.gnome.org/archives/gtk-perl-list/2000-February/msg00008.html

 - Greater customizability. (If anybody wants to go in and
   tweak the HTML, it's very straigthforward.)


Some questions I have:

 * Is the level of address mangling right? What it does is mangle:

     otaylor@redhat.com

   to:

     otaylor at redhat dot com

   Which won't stop determined spam harvesters, but might
   keep you from getting spam from stupid people ;-)

   It also doesn't put addresses in mailto: links to avoid
   making them too obvious.

   The two alternatives would be:

    - Don't mangle at all. This is what the pipermail archives
      did. If you are like me, you consider trying to 
      keep your address out of the hands of spammers useless
      task...

      The advantages are that searching works properly and
      mailto: links work.

      But I have had some (not too many, really) complaints
      about the current archives.

    - Aggressively mangle - e.g.:
 
        otaylor@xxxxxx.com

      So that neither people nor machine can figure it out. I
      find this quite annoying as an archive user, but it is the 
      only sure way of stopping harvesting.

 - Is the ordering in the thread index what people want? (The
   explanation at top is lifted from sources.cygnus.com.)

 - Is it distracting to have a search field at the top of each
   Thread/Date/Author index? Useful?

 - Anything else you would want in order to make the archive
   maximally convenient?


I should be able to move the public archives over to this new system
with just a bit of editing of config files and an overnight
run once I'm sure people are happy with it. 

Moving the private archives is more difficult, and probably
won't happen soon.

Regards,
                                        Owen