LISTSERV mailing list manager LISTSERV 15.5

Help for HTML-WG Archives


HTML-WG Archives

HTML-WG Archives


View:

Next Message | Previous Message
Next in Topic | Previous in Topic
Next by Same Author | Previous by Same Author
Chronologically | Most Recent First
Proportional Font | Monospaced Font

Options:

Join or Leave HTML-WG
Reply | Post New Message
Search Archives


Subject:

Re: Modular HTML DTD - Work in Progress

From:

"Daniel W. Connolly" <[log in to unmask]>

Reply-To:

[log in to unmask]

Date:

Thu, 11 Jan 96 01:24:34 EST

Content-Type:

text/plain

Parts/Attachments:

Parts/Attachments

text/plain (195 lines)




--part

In message <v02110103ad15f27f91bb@[192.188.119.193]>, Murray Altheim writes:
>>> http://ogopogo.nttc.edu/spec/html/modular/modular-dtd.html
>>>
>>I'm glad to see this released, in whatever form.
>>
>>It looks good, in general.
>
>Thanks. Still a lot to be done.

Sure. But now you'll get help! See:

http://www.w3.org/pub/WWW/MarkUp/html-pubtext/

My changes are attached below. Well... that was my first set
of changes. I've done another set to add INSERT and RESOURCE
since then...

>> * provide relative URLs as system identifiers, as well as
>> FPIs when linking from one piece of public text to another.
>> (for interoperability with stuff like SP, which groks HTTP
>> system identifiers)
>
>Isn't this handled in the SGML catalog file, external to the DTD? I'm under
>the impression that any system-dependent stuff should be external to the
>DTD, and that the SGML catalog file is where the mapping from PUBLIC
>external identifiers to system-dependent SOIs takes place. SGML Open's
>catalog definition does not provide for URL mappings, nor does the SGML
>spec.

Yes they do. My point is that we should look at the web as one great
by SGML system; then URIs are system identifiers. Check out the way
nsgmls handles http: URIs as system identifiers. Works great. Try this:

  nsgmls -m http://www.w3.org/pub/WWW/MarkUp/html-pubtext/catalog \
http://www.w3.org/pub/WWW/MarkUp/html-test/resource/spec-examples.html

When I finish putting in the relative URIs, we'll be able to just do:

  nsgmls http://www.w3.org/pub/WWW/MarkUp/html-test/resource/spec-examples.html

FPIs are only necessary if you want to change systems (e.g. from URI
space to the Mac file system).

> Don Stinchfield's paper "Using Catalogs and MIME to Exchange SGML
>Documents"[1] addresses this and provides proposed solutions for those
>systems that don't understand URL SOIs.

Sure... the catalog is handy for when you want to copy stuff out of
the web and use it in the local filesystem.

> It's a great feature of SP, but not
>part of any specification as I understand things.

It's just part of SP's entity manager. Since relationship between and
entity manager and a filesystem, diskette, network, chalkboard, or
typewriter is unspecified, any of those mechanisms can be used to
build a conforming parser.


>> * Can we move the SDA stuff to LINK declaratoins at this point?
>> Or do we have to keep them in the ATTLISTs?
>
>I was only following what was in the original 2.0/2.1 document. If you have
>specific recommendations as to what needs changing, I'd be quite happy to
>incorporate your ideas.

Actually, since you made the stuff available, I'm doing a little hacking myself.

> I'm frankly unclear as to how SDA entities are
>used, so I don't know what changes would be needed.

Me too. That's why I want them separated out -- so the SDA-knowledgeable
folks can maintain them.

>> * Get rid of %HTML.Recommended stuff

> Perhaps switching in and out certain modules would
>allow for both within the same DTD. I've been planning on doing that with
>'%HTML.Hilighting' anyway, so maybe 'HTML.Recommended' is another
>candidate. Just a lot more work...

Modules seems like a good idea: choose these modules for a permissive
DTD, and these alternatives for a clean DTD. Rather than marked sections
in html.dtd, just have html-edit.dtd and html-browse.dtd.

> BTW, maybe this is the time to ask: why
>was the semicolon left off of '%HTML.Recommended' when used as marked
>section status keyword ('%HTML.Recommended' vs. '%HTML.Recommended;')?

Because it's optional -- it's only necessary to disambguate, e.g.
%foo;bar


>>I'm curious about testing: I don't see any sample documents. Have you
>>done any testing?
>
>I'd just started installing nsgmls on a local system when I took a job in
>Cambridge, so I haven't had the time to complete and test.

Right.... I found a few typos...



>I'm not quite clear on this. Don't we really need a test of DTD
>equivalence, not document validation equivalence?

No. I don't want to reproduce the HTML 2.0 language. I want to find
out how the language has evolved -- how people are using it.

That's what the validation service was set up for -- so folks could
give feedback of the form "the DTD doesn't allow for this perfectly
reasonable looking document: ..."

>Documents produced by an SGML compliant editor would by nature
>validate against the DTD used to produce them, so it seems that your first
>idea (using those documents riddled with errors) would be better.

There are authoring tools out there now, but they're still not SGML-based.

Documents produced by non-SGML authoring tools are likely to be invalid,
but only in a finite number of ways that will be easy to catalog; contrast
this with the problem of cataloging the noisy errors that are made with
a text editor.


>In the content negotiation working group we're discussing HTML feature set
>negotiation. I'm considering changing the 'base' module name ('headings.*',
>'tables.*') to include ownerID ('ietf-html-v2.2-headings.*',
>'w3c-html-v2.0-ext-tables-v0.3.*'), so that there's a direct link between
>HTML extension and feature negotiation based on a keyword.

Take that to an extreme, and module names become global names -- i.e.
URIs. That's what I'm after: pick and choose modules from all over the
globe.


>Thanks very much again for your comments.

Thanks for the DTD work!

--part


Technical:
* added color module (body colors, font colors, ..
* added dynsrc module
* added font module
* added layout module
(hr size, image align, etc.)
@@ need to document how these are all obviated by stylesheets
@@ invent a macro notation to formalize it?

* removed some funky marked-section stuff (%body.forms, %HTML.Recommended, etc.)

* changed content model for table to accomodate some test data I found
(@@ need to review this change with Dave etc.)

* genent module: These are parameter entities!
added %URI
moved %just to layout.dtd

* Typos in comments in ISOlat1.sgm

* @@ my nsgmls doesn't have wide char support, so I had
to comment out much of the Unicode stuff

Editorial:
* got rid of all internal version lables (except
machine generated ones, using RCS keywords)
"status" is not a function of the info in
the file; it's a function of the world's perception
of that info. The status belongs in annotations
(ala PICS labels)

* Changed owner from IETF to W3C

TODO:

@@ draft "minimal HTML" DTD (level 0)
@@ draft editing interchange DTD
@@ draft stylesheet-happy DTD
@@ specify DSSL-STTP based macro mechanism to allow architectural
forms that express the notion that
<font size="+3">xxx</font> equates to
<span style="font-size: +3">...</span>
=============================================================================

--part--



Back to: Top of Message | Previous Page | Main HTML-WG Page

Permalink



LISTSERV.HEANET.IE

CataList Email List Search Powered by the LISTSERV Email List Manager