Re: Making the picons database distributed
Tue, 8 Sep 1998 10:08:06 +0100

On 08/09/98 09:38:04 Daniel Glazman wrote:

> In 1991, picons were only XBM and the compface algo was efficient enough
> to allow picon integration in X-Face. In 1992, XPM icons stopped that and
> the X-Face-URL has been proposed....and abandoned because it implies
> 1) HTTP implementation in MUAs
> 2) wait until you see the face of the sender
Well, 1's increasingly common anyway, what with the tendency to have active
http links in mail messages. But it doesn't need to be interpreted directly
by the MUA - xfaces could process it. It's just an alternative to designing
a server.

2 isn't *that* much of a problem, if the choice is between waiting while a
background process fetches a message, and not seeing a face at all. Again,
the advance of OS technology has made this more feasible.

However, Paul Harrington's suggestion for using LDAP is also a possibility
- but I suspect that there will be more resistance to a publicly-available
LDAP server at most sites, than to a web server. This is partially because
web servers are more understood, and partially because an LDAP server is
more likely to contain internal information too. An added aspect is that an
external interface to an LDAP server is most likely to be a web server

> This is not bizarre, this is impossible. Articles are thrown to /dev/null
> when they get too old ("too old" is a value set by the newsmaster).

I know; but it's possible to have a "downstream" site that is really just
your local faces server, so that you're using NNTP as the transport and
update mechanism, not the storage mechanism. But it's not really a sensible

> Futhermore, such a call for discussion will never ever reach a consensus
> allowing a call for vote !
True (and nor should it...)