Andy Dingley said:
What does "styled for web presentation using XSLT" mean ?
An XML file prepared for presentation through a web browser
(specifically, cutting out the stage of HTML conversion).
If I use XSLT on the server, I can use it to serve content to
everything from Telex machines and NS4 upwards.
But in what formats can the content be served? If the answer is
any format, as you suggest, then the contemporary content model
is misprised -- really, we seem to be dealing with an immanent
FO level within XSLT (i.e., the contemporary standards for Formatting
Objects are something like "cripple-ware" relative to rendering
technologies of the none-too-recent past -- why not Ghostscript FO?)
Do you mean "XML to the client, styled with client-side XSLT" alone ?
More or less (see above).
Yes - although at least one browser offered XSL in '99, it was a
rarity. It's still impractical to offer XSLT-alone content over the
web - most users are using IE 6, but there are plenty who aren't.
Only if you're some form of intranet or extranet where you can
influence browser choice can you really do this.
Well, obviously anyone can afford Mozilla and it has an XSLT engine.
I.e., the thought is obviously a little blue-sky but presentation is
going in certain directions and there are not necessarily overall
reasons determining this.
What does CSS have to do with anything ? You can express (some)
semantics with HTML 3.2 and liberal use of class and id attributes,
although this isn't enough to really allow more than simple end-to-end
communication between systems that are already hand-built to
understand each other.
Well, on my understanding <p> is a semantic marker: semantic structure
is
non-presentation structure.
CSS can be used to apply styling to a classified HTML markup, and it's
not hard to arrange the semantic description so it's also available to
drive the CSS. However the CSS selectors have no comprehension of
semantics, they're just using the markup coincidentally.
Well, CSS has to do with "latent" content as understood in traditional
layout. Have you ever had to do paste-up? You get the idea that what
you are sending "camera-ready" to the printers is not really that
arbitrary relative to established typographical standards.
The next generation is as much about communicable standards for
semantics, as they are about semantics themselves. I've been using RDF
since '99 for the internals of systems, but it's only recently that
DAML or OWL have made these semantics communicable to other systems
that haven't this tight and hand-coded link set up beforehand.
Incidentally, talking of RDF, then have a look at this week's new
document release.
Well, what is so fabulous about having a set-theoretic language (OWL)
handling textual content? The traditional understanding (deriving
primarily from Tarski) is, really nothing: if you were actually to do
semantically-based reasoning using a "calculus of inclusion" you'd run
into paradoxes, and a different kind of structuration is required to
do less stringent "reasoning" such as automated theorem-provers are
capable of (i.e., working recursions upon sets).
What does this mean ? XHTML certainly _wasn't_ designed with sole
reference to CSS (I did once have this very discussion with Dave
Raggett - should we ditch the lot in favour of <div> and <span> and a
bunch of CSS). XHTML exists in isolation. If it has simplistic
semantics, then that's for its own reasons - not because of CSS.
Well, you seem to be really locking with the W3 teleology here: XHTML
is a objectively defined standard with certain structural properties,
and the assertion was that one of these properties is tight
integration with CSS (i.e., there would be no need to do XHTML styling
using a more powerful styling language).
What's your obsession with CSS ? I'm beginning to wonder if you
have even a glimmer of a clue here. CSS lives at a presentation layer
- if you care about semantics, you need to fix this at a deeper level.
if you're to have any hope.
I'm beginning to wonder whether discussion of markup languages is not
allowed to be pitched at this level of abstraction anymore (i.e.,
whether I actually even have a thought to think about SGML and
Mostowski relative to the parameters specifying W3 consortium
standards). It appears that one is to instead have a "depth" model of
semantic markup, which I am saying doesn't make a lot of good logical
sense (i.e., what-all you said isn't just what you meant).
This is untrue. If PDF is predominant, that's just because PDF is a
popular target. It's not an XSL:FO limit.
Well, and it is too (i.e., it's designed for extremely static
presentation, *not even* ).
Well, I'd need to know more about what you
That's more because demand expands to fill available technology. What
are you specifically thinking of here, are are you just re-arranging
words you've overheard ?
I've never heard anybody say anything at all like this, actually.
Because semantic structure isn't really just "object-property"; it's
got all kinds of sides, and the minute you go to a relational level
with semantic content you have serious computational problems (it's
the *Entscheidungsproblem*, the unsolvability of which really the
fundament of computer science, in a different form). *That* is kind
of a commonplace.
And what does CERN have to do with things today ?
As little as possible (as far as I can tell).
Look at the huge amount of work going on right now with RDF and OWL.
Go to the Protege workshop in a few months. But no, you;ve already
decided that RDF "came a cropper"
Well, I'm not really in that business. But thank you for your
observations; I'll go familiarize myself with that RDF information.
P.S. By the way, who is Eric (not Tony)?