Fw: CSS and XSL

["Oren Ben-Kiki" <oren@xxxxxxxxxxxxx>]

Jelks Cabaniss <jelks@xxxxxxxx> wrote:

>I wrote:
>> Using one _syntax_ does not mean using one specific parser - and
definitely
>> doesn't mean using one monolithic program. Quite the opposite.
>
>Hmm.  I thought the big argument for XSL, etc. expressed in XML was to
parse
>everything with the same parser.  Are you saying The One Syntax is for the
>benefit of *humans*?


No. What I meant was that using XML syntax "for everything" means that it
would be possible to write many small reusable programs/modules/components,
each doing just one thing to an XML input and generating XML output. A
particular application would be built by combining an ad-hoc set of such
programs/modules/components.

It does not mean that necessarily the program would be one huge block of
code with an XML parser at the start, then doing all sorts of in-memory
stuff, then an XML emitter at the end. It might very well be a combination
of small programs each doing XML parsing, some processing, and XML emission,
in which case each program may use whatever parser it likes. Even if they
end using the same one, it would just be the equivalent of using <stdio> in
UNIX.

Even if you go the "single program" road, without going to/from textual XML
all the time (it is rather inefficient), you could still easily write (in
Java at least) an "XML shell" which would allow combining arbitrary SAX
filter modules (for example), in the same way that a UNIX shell allows
combining arbitrary ASCII stream programs.

If, however, some things are XML and some are CSS and some are in other
formats, this approach fails, since you can't apply an XML pretty printer,
XML validator, XML namespace handler, XSL XML-to-XML converter, XML-to-DB
and back converter, XML stream processor, etc. etc. to a CSS document. That
is, you can, but the CSS would be treated like a black box - just moved
around without any processing. You'd have to make a special case for it.

I feel that the above is a convincing reason to _allow_ an alternative XML
syntax for CSS (while fully preserving its current semantics!). You don't. I
also think that XSL should be the platform to do so (my "heretical view"),
and I guess most people on this list don't agree. I've heard two reasons:

- "This is what we started to do, we can't switch now" - pretty weak for a
standard whose 1.0 version is not final yet and which has only one
implementation (that I know of) of the relevant (FO) part.

- "CSS is bad for print" - so, what about fixing CSS?

Fine, that's what makes life interesting. Maybe Voyager group would be
interested? :-)

Have fun,

    Oren Ben-Kiki



 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list