This is the mail archive of the
mailing list for the DocBook project.
Re: docbook XML entities in nsgmls
- To: docbook at lists dot oasis-open dot org
- Subject: Re: DOCBOOK: docbook XML entities in nsgmls
- From: "Eve L. Maler" <elm at arbortext dot com>
- Date: Mon, 01 Nov 1999 12:59:05 -0500
- Reply-To: docbook at lists dot oasis-open dot org
Hi Bob-- ADEPT currently requires an SGML declaration because it is needed
to configure the parser that underlies the product, which supports XML
documents but was originally based on SGML. There is indeed a suitable
XML-friendly SGML declaration included; it should be in
.../entities/ati-xml.dcl, and in ADEPT 8.1 (I think), it gets called
silently when you indicate that your doctype application is XML- rather
than SGML-based. There is also .../entities/ati-gxml.dcl, which is used
for the ADEPT Global edition (full Unicode authoring support).
I had thought that nsgmls/SP never became fully XML compliant, which could
explain problems with (e.g.) hexadecimal numeric character references;
these simply aren't allowed in SGML unless your parser supports WebSGML,
which none really do.
At 09:25 AM 11/1/99 -0800, Bob Stayton wrote:
>Actually, I mentioned nsgmls because it exhibited the same
>symptoms as Arbortext Adept in this regard. My real
>problem was trying to use Arbortext with the Docbook 3 XML
>DTD. In the XML world, you officially don't need an SGML
>declaration, which is why it is not in the standard. So if
>Arbortext is claiming XML compatibility, why does Arbortext
>Document Architect 8 require an SGML declaration for an XML
>DTD? Probably because they haven't finished adapting DA to
>XML. OK, but if it requires a declaration, why doesn't it
>include one in the distribution? The product works quite
>well with XML, but they don't seem to make it easy.
>Norm, you must use Adept with Docbook XML. Didn't you have
>to dig up a declaration as well?
> > From Terry Allen:
> > It really doesn't matter what decl you use so long as it doesn't
> > trip you up. To detect unwanted Unicode characters would require
> > a much more finely tuned decl. And much as I like nsgmls (and use
> > it exclusively most of the time), to get all the little XML nits
> > you need to use an XML parser.
> > For example, I think nsgmls will not detect a comment that occurs
> > before the xml PI.