[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

RE: XSL intent survey


Subject: RE: XSL intent survey
From: Ed Nixon <ed.nixon@xxxxxxxxxxxxxxxxx>
Date: Fri, 20 Nov 1998 13:35:00 -0500

Here! Here!

Thankyou David Carlisle.

If there were a decent, comprehensive...book (yes) on the subject of DSSSL, I 
for one would be far less concerned that they do 'the right thing' with XSL. As 
it stands, DSSSL is a tool that seems to be used by sophisticates, academics, 
and specialists in the technical publishing industry: not, I don't think, 
because it is inherently difficult but because it has never been effectively 
supported with comprehensive learning material -- where is the 'Teach Yourself 
DSSSL in 8-1/2 Days!' tomb?

XSL, we can only hope will be able to survive the storms and stresses of the 
development process with a similar, if somewhat less comprehensive, set of 
cleanly integrated but differentiated features. With all due respect for 
surveys and questionnaires, it's not clear to me what the purpose or benefit 
will be other than to add more complications to what must be already a highly 
complicated, read: politicized, process. I'd feel much most positive about r  
esponding to something even semi-official from the Working Group. It's high 
time for a new draft, note, or whatever.

If action is needed, perhaps it should be to take Microsoft at it's word and 
respond to it's "Tell Us What You Think" link as well as support for some of 
the browser standards activities.

	...edN

-----Original Message-----
From:	David Carlisle [SMTP:davidc@xxxxxxxxx]
Sent:	Friday, November 20, 1998 12:17 PM
To:	xsl-list@xxxxxxxxxxxxxxxx
Cc:	xsl@xxxxxxxxxxxxx
Subject:	Re: XSL intent survey


> I probably don't know enough about
> document stylesheets to understand why you need the formatting objects
> and perhaps someone on the list can explain.

Because you want to specify the format of the output:-)

You are probably thinking in terms of HTML where there is some kind of
default formatting for a fixed set of elements (based on historical
actions of the browsers, mainly) and typically style sheets are used
just to tweak or fine tune those defaults.

For XML you don't have any default semantics, so if you want to say that
the content of a <p> element should be typeset as a paragraph you need
some way of referring to the abstract notion of a paragraph. That's what
the paragraph formatting object is. Similarly the other formatting
objects give the fixed set of `typographic' (may be rendered as speech
or braille or whatever) facilities that should be implemented.

It is rather sad that the `suggested' answers in this survey didn't
suggest the most obvious language on which to base this system (as it is
in fact the language on which large parts of XSL are based) namely
dsssl. (`base' can be taken rather loosely, for example people seem not
to like dsssl syntax (for (some (strange (reason))))).

DSSSL had two separate languages. A transformation language and a style
language. As far as I know the transformation part never got
implemented. The major reason why the non implementation of dsssls
transformation language has not turned out to be important was that
James Clark showed how one could effect transformations using the style
language (by having a set of flow objects that write SGML instances).
a dsssl flow object is more or less the same thing as an XSL formatting
object but in the current XSL draft this mechanism of using the same
paradigm to do both transformations and style is more cleanly expressed
than in dsssl (where it is a more or less non standard extension).

XSL has other problems (its transformations and patterns are not
powerful enough to say do anything with MathML), and it doesn't have many
of the needed formatting objects, but that does not worry me too much as
it is explicitly says in the draft that more will be added later.

Actually I can not see any use at all for having a style language that
can not do transformations. Given that you are going to have to
transform the original parse tree to produce the required output tree
for the formatting, why not just add the formatting characteristics as
you construct that tree (as you can do in a combined language) rather
than have to go over the tree again adding formatting characteristics
with a second style language?

David


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


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



Current Thread
Keywords