David Herron (david@TWG.COM)
Mon, 24 Jun 91 18:53:06 -0700

Just heard 'bout this project and just finished glancing through the
source & saved discussion. Very interesting. It was especially
surprising while reading the source & finding a couple of files with my
name in the copyright notice! Especially since, as I recall, they
weren't copyrighted when posted to comp.sources.misc!! :-)

Ohwell.. nomatter..

A couple of points ...

I don't see the purpose to writing a distributed face server. It seems
that writing distributed databases is a tough job. It doesn't make
sense to me to have umpty ump specialized distributed databases for
all the different kinds of things which people want to distribute.

Adding X-Face: lines to e-mail & news messages seems like the
best idea to me..

I don't see any documentation of the format. I see this and, since I'm
writing an e-mail UA right now, want to add this capability into my
UA. But all the source is copyrighted in a way the company wouldn't
like, after all we'd certainly want to recieve `moneys' for the UA ...
So this means read the code & reimplement, eh? Not too big a deal (maybe).

This is a cool facility & I'd like to see it get widely implemented.
My impression so far is that it is a nice implementation too. The
picture size (48x48) should be large enough and monochrome lacks some
sizzle, but adding enough colors to be useful also brings the data
requirement large enough to be worrisome. An extra 2-3 lines of text
from the current X-Face: stuff is a trivial cost for what it gains

But why put it in the header? This isn't data which is used for
routing messages around. On the other hand it is something which is
used by a UA while displaying the message .. one of the ideas toying
about my mind right now is a folder-browser which has the X-Face:
picture next to each `scan line'. So perhaps it belongs in the header
after all. Putting it in the body, as it's own body part (see X.400
and the recent multimedia mail IETF draft), might be a layering
violation. That is, it would place some semantics on the body parts
which I don't recall seeing in any of the standards documents.

Seeing as there is an IETF working group working on RFC-822 extensions
isn't this a good time to bring X-Face: to their attention?

Hmm... seems X-Face must use some data compression. The raw data format
would require 2304 bits or 288 bytes. Encoding this will increase it
out to somewhere around 432 characters, which would require 5.4 lines
of 80-column text to display.

Does the data identify attributes like width & height & number of
bit-planes? If not then it should, so that the format can be extended
later. With the caveat that kiddies might start putting
nekkid wimmen in their X-Face: lines if they have something which'd
support lots of picture formats... ;-)

Have fun,