[standards-jig] On the Subject of Avatars
bob at wyman.us
Sun Jul 20 21:55:29 UTC 2003
Evan Prodromou wrote:
> I dislike JEP 0084 for the following reasons:...
> * requires pubsub
The use of PubSub overcomes what would, for many people,
be a major objection to the approach that you propose. That
is: Your approach requires polling of the vCard from time to
time to determine whether or not the avatar has changed.
Given that avatars, as images, can be "large", this means
that it is likely that a great deal of network bandwidth will
be consumed in regularly discovering that the avatar has not,
in fact, actually changed.
Admittedly, requiring something to use PubSub today, when
it is still not an accepted standard and there is very little
experience with it, can be seen as a negative. However, in
time, as we all become more familiar and comfortable with
PubSub, the situation will be different. Personally, I think
that when one writes a standard, you should be focused on the
world as it will be as the standard will be used -- i.e. the
world of the future -- not the world as it exists today.
Today, PubSub is new, unfamiliar and to some a bit strange.
In the future, it will be common, familiar and the obvious
solution for avoiding polling in many, many applications.
I wonder if a "compromise" might be appropriate here. For
instance, how about proposing that the vCard carry just
the "meta-data" that JEP-0084 defines? This would reduce the
complexity somewhat without burdening the vCard with the
weight of the avatar images. On the other hand, we must
recognize that the meta-data changes from time to time, as
can any of the content of the vCard. Perhaps what we need is
a PubSub node that one can subscribe to in order to rapidly
discover when a vCard has been changed...
More information about the Standards