Hi Jon,
A command line can reference more than one XML catalog, so you could add
your own catalog for the entities. But if you cannot alter the processing
command or catalogs, then shipping the entity files and using local
references will certainly work and be simple enough to debug.
To simplify the DOCTYPE of your documents, I would suggest creating a single
system entity file that references all the iso entity files with relative
file references. That way your DOCTYPE only need to refer to a single file.
Bob Stayton
Sagehill Enterprises
bobs@sagehill.net--------------------------------------------------
From: "Jon Leech" <
jon@alumni.caltech.edu>
Sent: Thursday, October 31, 2013 11:28 PM
To: <
docbook-apps@lists.oasis-open.org>
Subject: [docbook-apps] Catalog setup issues in using MathML entity sets
(was Re: Should xmllint successfully validate docbook 5 containing
XIncludes?)
> On 10/31/2013 09:36 AM, Bob Stayton wrote:
> > Hi Jon,
> > I think Carlos meant to use the official MathML namespace instead of a
> > DTD reference:
> >
> > xmlns:mathml="http://www.w3.org/1998/Math/MathML"
> >
> > It seems loading the MathML DTD confused Jing. The DocBook 5 RelaxNG
> > schema permits elements in the MathML namespace, but it does not try to
> > validate them. But it seems your problem is recognizing the named
> > character entities in the MathML DTD. If Jing does not like loading the
> > entire MathML DTD as you tried, it would probably accept loading just
> > the entity declarations from the DTD. You could try copying the MathML
> > dtd to another file and stripping out all *but* the entity declarations.
> > Those are conveniently located at the end of the DTD file. Then
> > reference that file in the parameter entity in the DOCTYPE to load those
> > entities.
> >
> > Then Jing should validate against the DocBook 5 RNG schema. If you
> > also want to validate the embedded MathML elements, you will need a
> > validator compatible with NVDL such as oNVDL (
> >
http://www.oxygenxml.com/onvdl.html ).
>
> Thanks very much Bob, and Carlos. This is very helpful. If I load
> just the four entity declarations I require, jing validates my DB5
> source without hanging.
>
> However, I then ran into an issue with the catalog setup. My
> system's XML catalog (which includes whatever the Debian docbook and
> mathml packages I have installed put there) contains the identifiers for
> mathml2.dtd. Loading that also pulls in the co-resident local copies of
> the entity sets I need from
>
>
http://www.w3.org/TR/xml-entity-names/#sets>
> But I'm no longer loading mathml2.dtd due to the aforementioned
> problems with jing, and neither the public nor system identifiers
> corresponding to the specific entity sets I require are in my XML
> catalog. So when I try to include them individually in my documents,
> with a doctype prolog such as
>
> > > "-//W3C//ENTITIES Added Math Symbols: Delimiters//EN//XML"
> "http://www.w3.org/2003/entities/2007/isoamsc.ent"
> >
> ]>
>
> they aren't found (unless I allow the processor to fetch them from the
> net, which is of course very slow).
>
> Short of tweaking the system XML catalog setup, which is not
> something I can expect all of the people who will be using my documents
> to understand or want to do, distributing copies of the entity sets
> bundled with my Docbook source seems like the simplest way forward.
>
> Would you agree, or am I missing some straightforward way (not
> requiring root access, not OS-specific) to make XML lookups for these
> specific entity sets check in the same place as mathml2.dtd, even
> though their identifiers aren't in the system XML catalog?
>
> Jon Leech
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
docbook-apps-unsubscribe@lists.oasis-open.org> For additional commands, e-mail:
docbook-apps-help@lists.oasis-open.org>
>
>