Some comments, though. (Not as many as last time :-)
o The ALIGN attribute for TAB takes "decimal", and there is a DP
attribute (should this be specified to default to "."?) but the
descriptive text only talks about "Named left, center and right
tab stops".
o TABLE, TR, TH, and TD also take a DP attribute, but their ALIGN
attributes take %horiz.align, which does not include "decimal".
Nor is there any way to specify a decimal-aligned field in the
table's COLSPEC. I would suggest adding "decimal" to %horiz.align
and "Dnn" to COLSPEC. (Perhaps "D.nn" if you want different rows
to have different default alignment characters.)
o I was surprised at the break between <A> for footnote marks and
<FN> for footnote bodies. In retrospect it's a very nice idea.
The only potential problem I see is for browsers that wish to
treat footnotes differently from other anchors. They will not be
able to know that an anchor points to a footnote until the body is
seen. It may also confuse users, who are used to assuming that
when a client presents a link pointing to "#tag" that they will
eventually get to the referenced text by scrolling.
o I understand the motivation for changing the ROLE attribute on
NOTE to CLASS. I am afraid, however, that in the absence of some
standard set of names, people will not be able to customize
browsers to substitute icons. This will be especially true if
people name their classes in the local language. (I may have my
browser to know about "DANGER", but how about "PELIGRO"?) Perhaps
the DTD should have a section of standardized class names, not
only for NOTES, but also for DIV, EM, UL, and OL. This would
leave people free to add their own, but would make it more likely
that they would choose the same name for similar concepts.
o The spec seems to say that the LANG attribute must include a
country as well as the language. While I can see that it is be
useful to distinguish "pt.br" from "pt.pt", I'm not sure why an
author should be forced to make the distinction rather than being
allowed to merely specify "pt". Clearly, in most cases, users
will want to configure their browsers based on the language alone
rather than the specific regional variants.
o In the SHAPE attribute, it says that the coordinates are
"interpreted as fractional width/height". Presumably this applies
to the radius, width, and height components as well. For circles,
this would appear to be ambiguous. Is it fraction of width,
height, smallest?
o Since there is no way to put markup in IMG ALT, it would be nice
if FIG had an ALIGN=inline value.
----
Evan Kirshenbaum +------------------------------------
HP Laboratories |K: Natives. They may be hostile
1500 Page Mill Road, Building 4A |C: Well, we're all a little hostile
Palo Alto, CA 94304 | now and then. Some of us are able
| to sublimate. Others just can't
kirshenbaum@hpl.hp.com | adjust.
(415)857-7572