Re: proposed new tag: IMG

Marc Andreessen (
Sat, 13 Mar 93 05:18:03 -0800 writes:
> >We're not prepared to support INCLUDE/EMBED at this point; it raises a
> >number of nasty issues that are quite separate from the idea of
> >inlined images. For example, what happens if one EMBEDS a document
> >that in turn EMBEDS the first document? Oops.
> I would consider this an error of the author that needs to be detected
> to protect the browser. It only requires maintaining a stack of
> nested EMBEDS.

Still, I'd argue the principle is still there. If a document has to
be pieced together on the fly, it could get arbitrarily complex, and
even if that were limited, we'd certainly start experiencing major
hits on performance for documents structured in this way. This
essentially throws the single-hop principle of WWW out the door (well,
IMG does that too, but for a very specific reason and in a very
limited sense) --- are we sure we want to do that?

> Some other hypertext systems do this, in a sense: in Guide there are,
> apart from real GOTO stype hyperlinks, also "folds" (I think they are
> called) which are sort of embedded documents that you can open and
> close. The advantage in certain situations is that opening a fold
> retains more context than following a link. It feels like using an
> outline processor, which is rather pleasant (for certain kinds of
> information).

Would be nice to be able to do this, but it would essentially be just
a fancy multi-windowed version of the current distributed hypertext
model, right? Instead of jumping to whole new documents, the act of
triggering a link would open up more information in the current
context. That doesn't require EMBED functionality...

> PS. I keep hearing about supporting MIME. Is there consensus on the
> form this should take? (I remeber getting in an unpleasant fight with
> Dan Connolly about this once -- BTW is he still with us?)

Mosaic currently recognizes documents suffixed .mime and throws 'em
off to metamail. Some support for MIME is in Tim's new WWW library
(although I haven't looked very closely at it). Other than that, I'm
not sure what the consensus exactly is. And Dano's history, I do


Marc Andreessen
Software Development Group
National Center for Supercomputing Applications