>Brent Baccala <baccala@...> writes:
>> Anyway, I've come across some documents that don't parse right. They
>> appear to have been generated by Microsoft Office, and include tags like
>> this:
>>
>> <![if !supportEmptyParas]>&nbsp;<![endif]>
>[...]
>I don't know SGML well enough to tell if this is something worth
>supporting or if this stuff is valid SGML at all. Does anybody else
>know?

I've been trying to read /The SGML Handbook/ (containing the complete text
of the standard) for a while now, and that's sort of like trying to sit
down and /read/ a circuit board. But I think I've /seen/ every page of it,
at some time or other; and I'm fairly certain I've never seen that
construct -- because, if nothing else, it's redundant next to marked sections.

And incidentally, even marked sections are listed at
http://www.w3.org/TR/html4/appendix/notes.html#h-B.3.3
as among the SGML constructs (including SHORTTAG things like <name/.../,
<foo<bar>, <>, and </>) as things you noone should expect HTML parsers to
understand.
They add: "We recommend that authors avoid using all of these features."
Pretty strong words, considering the W3's usual delusional mind-set about,
well, nearly anything to do with user-agents. ("You mean Netscape doesn't
have nsgmls in its parser?!?")

(And normally I'd take this opportunity to say something really mean and
possibly libellous about Microsoft, the SGML standard and its authors, and
maybe ISO, but I've just woken up, so you'll have to make do with a general
"Feh!")


--
Sean M. Burke sburke@... http://www.spinn.net/~sburke/