These bitmaps would be known to be available to document authors,
and non-graphical clients could possibly use a textual supplement
(e.g. "[LEFT]" for leftarrow-large, etc.). The bitmaps wouldn't
need to necessarily look exactly the same from client to client.
(b) An optional parameter for <LI> to allow specification of a pointer
to an image to inline (rather than using a bullet or whatever).
This would let one do something like:
<LI IMG="internal-gopher-menu"> A Gopher menu.
<LI IMG="internal-gopher-image"> A weathermap.
... and then one would get a slick bitmap-enumerated list. Again,
text-based clients could substitute the normal list style or
textual equivalents (e.g., "[menu]" and "[image]").