| My interpretation of things (despite 4.283, 6.2.3, and production),
| is that the lack of an explicit SGML declaration does not prevent an
| entity from being an SGML document entity. That is, an explicit SGML decl
| is effectively optional.
I'll agree with "effectively optional," but that's a different story.
The only normative definition of "SGML document entity" is that given
in 8879. As in other cases, 8879 fails to name a useful thing, a
document with a doctype decl but no SGML decl.
| For example, I would say that an HTML document with an explicit doctype
| declaration as specified in RFC 1866 is a valid SGML document entity. An
But if you're using the language of 8879, you'd just be wrong. And
you don't have to hold that view re HTML because RFC 1866 doesn't use "SGML
document entity," so far as I can detect.
| SGML system (be it the entity manager part or the TR9401:1995 processing
| part--assuming that isn't the entity manager anyway--or whatever) could
| either have an implied SGML decl that will work with the document (i.e.,
| have an internal default implied SGML decl that is "big enough") or it
| could intuit an implied SGML decl that will work (e.g., by finding a
| DTDDECL entry in some TR9401 catalog that it may have access to).
Yes, but that's not relevant to the matching up of terminology.
| Certainly, the CALS folks have almost *never* interchanged SGML decls even
| though MIL-M-28001 specifies an SGML decl that is not the Minimal or Basic
| one defined in 8879, but I think that saying that all CALS documents are
| not SGML documents is just not fruitful. Perhaps in some interpretation
| of 8879, you might have a point, but I bet that, if push came to shove,
| and you asked ISO/IEC JTC1 SC18/WG8 for a ruling, you'd find they(we)
| decide to allow an entity without an explicit SGML declaration to be
| considered an SGML document entity. (However, I cannot speak for WG8.)
Even if I asked ISO/IEC JTC1 SC18/WG8 for a ruling today, how would that
change the plain meaning of the text of ISO 8879? If WG8 wants to
*revise* 8879 along those lines, fine, and after the revision is ballotted
and adopted by ISO, then things would be different.
We don't need to reinterpret 8879 to deal with the matter, though. One
need not provide an SGML document entity to an application (HTML on
the Internet) when an SGML declaration has been specified, as for HTML.
The question about prepending an SGML declaration was asked in the general
case, and that's how I answered it. My answer came out differently than
Dan's because I was considering a wider range of cases.
Terry Allen ([log in to unmask]), Online Books Editor, Songline Studios
affiliated with O'Reilly & Associates, Inc.
A Davenport Group sponsor. See http://www.ora.com/davenport/README.html
"Laid across a map of the US, Indonesia would stretch from coast to coast."