Mozilla XML - Links in DocBook XML

Author

Links in DocBook XML

Steven T. Hatton

2007-05-12, 7:12 pm

I have been carrying on a discussion with some folks on the DocBook mailing
list regarding the use of XML and CSS, without any XSLT transformation to
XHTML. I believe the W3C has recommendations for doing these types of
things, but I really don't know the state of the implementations.

The whole discussion is archived here:

http://lists.oasis-open.org/archive...1/msg00019.html

I am very busy right now, and really don't have time to follow up on this
topic by researching the W3C recommendation and specifications.
Nonetheless, I believe it is very important. My assertion is basically
that, in principle, it should be relatively easy to serve out XML documents
without transforming them to XHTML, and you can still have links which
reference other parts of the document, or other URLs external to the
document. The document can be served in smaller sections so that the
client does not need to transfer the entire document (read book) just to
view a section of a chapter.

One of the primary objections is that links cannot be effectively supported
in XML. Could someone please address this issue? I really want a W3C
compliant solution as opposed to a Mozilla specific one if at all possible.
The issue of whether the solution works with other browsers is important,
but secondary to my primary objective which is proof of concept.

--
STH

Noah Slater

2007-05-12, 7:12 pm

The thread is split, see
http://lists.oasis-open.org/archive...eads.html#00036
for more

Peter Flynn

2007-05-12, 7:12 pm

Steven T. Hatton wrote:

> I have been carrying on a discussion with some folks on the DocBook
> mailing list regarding the use of XML and CSS, without any XSLT
> transformation to XHTML.

Basically, don't waste your time. Do it all server side and send
your users HTML. No browsers implement all of CSS2 yet, and none
of them implement all of XML yet either, so for the moment it's all
in pieces all over the floor, and will break for anything other than
a trivial document.

I'm uncertain why this is. Browser-side realtime parsing and styling
of SGML was possible nearly a decade ago, at acceptable speeds on the
slow lines we had then, using a lightning-fast 120MHz PC running W95,
Netscape, and a copy of Panorama. This retrieved full DTDs and style
sheets, with full entity resolution, and performed a tokenized parse
omitting only some aspects of validation -- more than adequate for the
job. It implemented bidirectional and multi-headed HyTime links, and
a load of other stuff that browsers have been promising for years.

Quite how Netscape, Mozilla, IE, Opera, and the rest have dropped the
ball so badly is not clear, but it is utterly inexcusable that ten
years after the technology was in daily production use as a plugin,
we still don't have anything even remotely near it in a production
browser (with the exception of DocZilla, of course, which is based
on the same technology [Synex] used in Panorama).

///Peter
--
XML FAQ: http://xml.silmaril.ie/

Christian Biesinger

2007-05-12, 7:12 pm

Steven T. Hatton wrote:
> One of the primary objections is that links cannot be effectively supported
> in XML.

What's wrong with XLink?

Simo Melenius

2007-05-12, 7:13 pm

Peter Flynn wrote:

> Steven T. Hatton wrote:
....[color=darkred]
> Quite how Netscape, Mozilla, IE, Opera, and the rest have dropped the
> ball so badly is not clear, but it is utterly inexcusable that ten
> years after the technology was in daily production use as a plugin,
> we still don't have anything even remotely near it in a production
> browser (with the exception of DocZilla, of course, which is based
> on the same technology [Synex] used in Panorama).

Do you mind if I bring up a correction here, albeit being a bit late :-)

The Synex thing has confused people before, too. That's probably because of
Citec's past with Synex and its Synex-based browser, MultiDoc Pro. However, it
was exactly because Synex disappeared from the market why DocZilla is _not_
based on Synex technology. Hence, Citec had to build another browser based on
another technology to support their SGML browser option, and the decision was
to start from the Mozilla codebase.

Currently, DocZilla is pretty much Mozilla Navigator with SGMLish twists. It
uses the standard Mozilla Gecko layout engine to display SGML using CSS (or
CSS2-ish, as implemented by Mozilla.org) as the styling language. But to load
and parse SGML, it embeds OpenSP (the reincarnation of J. Clark's SP parser,
used in nsgmls) which is a strictly validating SGML parser with full support
for DTDs and entity resolution through catalogs. Together with a few SGML/XML
specific features, like support for commonly used subset of HyTime, you can
probably call it a true SGML browser despite the fact that it isn't written in
the 90's and not based on Synex :-)

* * *

ObTopic: You can do live DocBook easily with DocZilla, including links. In
fact, the online reference manual for DocZilla is just one big collated
DocBook XML file which is simply viewed in a new window when the user selects
it in the "Help" menu. We turn DocBook's ULINKs (URL references) and LINKs (ID
references) into real hyperlinks using HyTime clinks. DocBook allows insertion
of the critical HyTime attributes into the DTD simply with a few helpful
parameter entity definitions. There's no hard-coded DocBook support in
DocZilla, as we can do it by parameterization.

br,
S

--
simo.melenius@citec.fi System Designer / Unix Applications
CiTEC Information, Service Development

Peter Flynn

2007-05-12, 7:13 pm

Simo Melenius wrote:

> Peter Flynn wrote:
>
> ...
>
> Do you mind if I bring up a correction here, albeit being a bit late
> :-)
>
> The Synex thing has confused people before, too. That's probably
> because of Citec's past with Synex and its Synex-based browser,
> MultiDoc Pro. However, it was exactly because Synex disappeared from
> the market why DocZilla is _not_ based on Synex technology. Hence,
> Citec had to build another browser based on another technology to
> support their SGML browser option, and the decision was to start from
> the Mozilla codebase.

I do apologise: I thought DocZilla supported the Synex .entityrc and
stylesheets.

> Currently, DocZilla is pretty much Mozilla Navigator with SGMLish
> twists. It uses the standard Mozilla Gecko layout engine to display
> SGML using CSS (or CSS2-ish, as implemented by Mozilla.org) as the
> styling language. But to load and parse SGML, it embeds OpenSP (the
> reincarnation of J. Clark's SP parser, used in nsgmls) which is a
> strictly validating SGML parser with full support for DTDs and entity
> resolution through catalogs. Together with a few SGML/XML specific
> features, like support for commonly used subset of HyTime, you can
> probably call it a true SGML browser despite the fact that it isn't
> written in the 90's and not based on Synex :-)

Thanks for the clarification, and full marks to DocZilla for doing it
right.

(Still doesn't excuse the other browser-makers for failing to make
the grade, though :-)

///Peter

Simo Melenius

2007-05-12, 7:13 pm

Peter Flynn wrote:

> Simo Melenius wrote:
> I do apologise: I thought DocZilla supported the Synex .entityrc and
> stylesheets.

OK, a couple of more FYIs for everyone:

Entityrcs, as a concept, _are_ in fact still alive: we decided that they are a
pretty good way of assigning auxiliary (to SGML and catalogs) properties to
documents pertaining to a certain doctype or public ID so we made DocZilla
just reuse the concept. Therefore, while there are some subtle differences the
Synex entityrcs and DocZilla entityrcs look pretty similar as well, both
syntax-wise and keyword-wise.

On the other hand, the SSH stylesheets are not practically supported since
Gecko is completely factored around CSS. There's still some, albeit very
indirect, support for SSH though: you can convert from SSH to CSS using a
simple transformer program we provide on the DocZilla website. It's a very
simple tool and the author still needs to clean up the resulting CSS file
somewhat but it might help half-way in converting a long SSH stylesheet.

br,
S

--
simo.melenius@citec.fi System Designer / Unix Applications
Tel: +358 50 347 8373 CiTEC Information, Service Development