Making the picons database distributed

T.J.A. Thurman (tjat2@cam.ac.uk)
Fri, 4 Sep 1998 19:05:40 +0100 (BST)

Just an idea I was turning over and wondered what people thought...

It could be worthwhile to store (parts of) the picons DB in some
distributed format. Remember that many years ago the Internet's host names
table was stored in a standard text file, which grew into the distributed
DNS we have now? Similarly, it might be useful to have the option of
keeping picons for a particular site on a machine at that site.

The WWW may as well be harnessed to do this. The picon for the user
foo@bar.com could be stored, for example, at
http://www.bar.com/0picons/users/com/bar/foo/face.gif
and the "unknown" picon for the same domain would then be found at
http://www.bar.com/0picons/domains/com/bar/unknown/face.gif

Note:
1. I suggested "0picons" because that name is unlikely to be used anywhere
else ("picons" could be a directory discussing picons as a subject, for
example).
2. The reduplication of the address in the URL allows sites which already
have a picons database merely to symlink "0picons" to /usr/lib/picons.
(Such a site could, of course, also be used as a local mirror of the
whole database for local machines.)
3. It might be more sensible to remove the "users"/"domains" part of the
address. However, this may conflict with the symlink benefits mentioned
above.
4. Naming the host "www" would save trouble in setting up. It would have
been elegant to standardise on a name such as "picons", but this might
delay acceptance.
5. Of the three formats widely used for picons, it may be best to
standardise on GIF for these purposes. Only GIF and XPM can subsume
the other two, and XPM is not widely supported by browsers (AFAIK).

I'd suggest the order of looking for a picon would then be as follows:
1. check local databases, as discussed in the Picons FAQ
2. check the local site, as above
3. check the Picons Search Engine.

Naturally, this sort of thing needs to exist in widely available mailers
in order to gain widespread acceptance. The availability of the source for
Netscape, and mozilla.org's acceptance of improvements, provides an
interesting opportunity for this.

Any thoughts?

Thomas

Thomas Thurman, Sidney Sussex College, | Any attempt to brew coffee with a
Cambridge. CB2 3HU | tjat2@cam.ac.uk | teapot should result in the error
http://www.sid.cam.ac.uk/~tjat2/ | ><> | code "418 I'm a teapot". --RFC2324