Andreas Prilop <nh******@rrzn-user.uni-hannover.de> writes:
I just noticed that documents with – and the like
http://www.unics.uni-hannover.de/nht...sgml-char.html
are again (!) valid HTML 4.
Being to young (well, in terms of Internet mangling, which doesn't
reflect my age compared to acquaintances :) to know otherwise, when and
how was that different?
An SGML parser should treat a character reference as data at the point
of its appearance; the unused codepoint, in return, is the *result* of
parsing that data (and the problem of the application :).
'Non-SGML-character' in terms of the SGML declaration is significant to
*markup* (such as entities, which are commonly confused with character
references just because /Entity Reference Open/ ['&'] and /Character
Reference Open/ ['&#'] share a leading character -- an unused codepoint
which is *subject to parsing*, like referencing an entity declaration
which contains a numeric reference to such a codepoint (and isn't
declared to be CDATA), on the other hand, should flag an error).
--
"On Usenet, parodies fail, because Usenet itself is a parody."
--Korpela's 13th Law about Usenet
