The alternative is to take IMG out of level 0. I suppose that matches
current practice, in that folks that want to reach text-based
consumers actually maintain separate document trees.
Hmmm... this is exactly the kind of thing I want to be able to
automate. Currently, the Commercenet stuff says "Click _here_ for a
text-based version of our documents." Their server should be
_able_ to detect the capapilities of the browser and deliver
appropriate documents, without the intervention of the user. That's
the whole point of format negociation.
This is really getting nasty: what level of browser in lynx? It does
forms and highlighting, but it doesn't do images. This suggests that in
stead of:
Accept: text/html; level=2
the more appropriate design is:
Accept: text/html; highlighting=yes; forms=yes; images=no
This starts to look like more cost than benefits. Hmmm...
So how do we gracefully deploy changes in HTML?
Dan