At 20:10 01/12/2003, Karsten Wade wrote:
>The target file in the common directory is called via a relative
> >i.e. it's always in an expected location of off the Fedora docs cvsroot,
> Are all authors expected to work with that relative path?
I am suggesting one idea for this, and yes under this idea, all authors
would work with that relative path. To work on a document, you would
need to check out the common directories module (of at least one top
level directory), and put it in a specific relative path.
The downside of which is that it may not suite all our
This idea is not yet a procedure; we are discussing that. I actually
like the idea of using a URI, will it work within the entity?
Yes, since its the parser that expands the entity.
... Any reason to use an entity though?
seems to fit into the spirit of things. :) It's structurally different
from a CVS module on a relative path, but it's essentially the same
thing - a remote canonical source you generally only have read access to
and you keep a copy of locally.
Rationale (for some),
We may not always be connected to the net.
Hence, for example,
<!--This group handles all docbook items. -->
<!-- Use formal public identifiers in preference to SYSTEM identifiers -->
<!-- docbook 4.2 -->
<public publicId="-//OASIS//DTD DocBook XML V4.2//EN"
Is a snippet of my docbook stuff.
The Edited text refers to the web,
the catalog maps it to /sgml/nw/docbook/docbookx.dtd
(Similar, using Oasis catalog, for emacs use)
2. Common Entities in DTD (please correct my inaccuracies)
Depends on the design modularity you want/need.
Some things (names etc) will be very stable.
Versions may change fairly quickly.
A. Common entities are added to Fedora specific DTD and served from
B. Authors calling the DTD gain access to the entities, which are listed
XXXXX (manually on the FDP pages? extracted via xml-fu from the DTD?)
Don't care? If they nest nicely they'll all be accessible,
either from the redhat site, or locally when mapped.
Two (local) usages,
Within the editor for use (not always needed, but useful)
Externally, when validating, say with nsgmls,
there the catalog is needed, so perhaps some documentation
on using Daniel Vaillard's software as a parser?
I know he supports catalogs.
C. Local copies of the DTD are kept, as is common.
I like it. Its flexible. redhat.com
has the masters,
they are available over the net if needed,
I can copy them to my hard drive if needed.
I think that meets most needs.