- From: len bullard <cbullard@hiwaay.net>
- Date: Wed, 11 Jun 1997 22:19:58 -0500
- To: Joe English <jenglish@crl.com>
- CC: w3c-sgml-wg@w3.org
Joe English wrote: > That was the original intent as I understand it, > but I've never seen <!NOTATION...> actually used that way, > and it would make very little sense to do so, at least > for general SGML. Actually, that is precisely how the MID specification indicates a "known" processor which the preparer of the document is indicating will work. It was convenient. > The idea that there is *one* application capable of doing all > possible processing on a data entity, and moreover that the > location of that application should be hardwired into the DTD, > is in direct contradiction to the principles of data reuse and > system independence. Hmm. It is an odd place to put it, I grant you. Would you preferred it to have been in the instance? If you wanted to move the information/data naming the processor or processor type, and you don't have MIME, how would you do it? > The situation might be different for XML; since we're assuming > a Web-based environment it *might* make sense to specify the URI > of a Java applet -- all that's missing is an API with which > the applet can interact with the application -- but that would > leave Perl, Python, Tcl, et cetera-based XML processors out of > the picture. IMO Internet Media Types (MIME types) would make > much more sense as notations' SYSTEM identifiers in XML. It indicates the type of a processor/handler. End of story. How does that get registered? IOW, VRML/JAVA have CLASSPATH: will XML have some equivalent for that? > > If my understanding of NOTATION is correct (from section 4.6 in the > > working draft at > > http://www.textuality.com/sgml-erb/WD-xml-lang.html#sec4.6), then a > > NOTATION in effect says "I don't know how to deal with this external > > resource, but here is the identifier of some processor that does." Data > > types would be saying "This is a number (date, time, etc.) I make no > > recommendations regarding processing." > > The XML draft does lend itself to that interpretation. > In full SGML though -- where NOTATIONs can be associated with > external CDATA entities (which XML lacks) and elements -- > they are often used to indicate data types in the sense > you describe. They do both, but they don't indicate data types, but name a document that defines the data types. Correct? So in that sense, they are in situ, only names. One is the name of a document; the other is presumably the name of a processor that the DTD emitter *says* handles whatever that document defines. How useful is that pair of names in that form? len
Received on Wednesday, 11 June 1997 23:24:48 UTC