xml-dev - Re: FOs again

[Rick JELLIFFE <ricko@geotempo.com>]

Amy Lewis wrote:
> 
> Following the discussion of FOs here, I thought that I needed to know
> more, so I went off and reread the requirements doc and the current
> spec.  That made me unhappy.  The requirements doc reads rather like
> a laundry list; the spec in turn rather like a receipt ("okay, did
> that one, what's next?").  That is, there seems to be no overall
> vision, or integrity, to the documents, so they go on and on and on.
...
> Summary: keep it simple, finish it soon, and focus on display on
> electronic devices with default behavior for other display types;
> make it extensible to provide finer control both on display (as
> desired) and on other devices.
 
CSS already provides this approach: there is no way that XSL could
succeed if if tried to be an alternative to CSS.

CSS starts with simple, declarative, cascading, display-oriented
features, allows extensions, and is gradually having more extensions
standardized. Already it has had to have one major revision to its basic
model, as it has moved from being a toy language.

XSL starts from the opposite approach: the people involved have been
involved in professional publishing or in the development of other
formatters. The reason that XSLFO needs to pay great attention to
figuring out what professional quality requires is so that quality
systems that non-professionals can use can be built.

Microsoft Word is the number one word processor in every major market
except Korea, apparantly. It has been localized for so many different
scripts and languages. For XSL-FO to succeed it has to provide some step
forwards for the W3C members (Adobe, Interleaf, Arbortext, INSO, even
Microsoft) in analysis, elegance and coverage: I would guess one could
say that for each vendor XSL-FO must provide a convenient step forwards
compared to their existing products *and* support at least as many
features as MS Word can.   To the extent that this might be true (is
that enough equivocation?) it would represent a consistent vision. 
(And, I would add, that for at least one non-commercial member of W3C,
the requirement for XSL-FO is that it be be able to support open source
implementations which provide better support for Chinese than the
20-year-old design of TeX or the 30 year year old design of troff.) 

The fly in the ointment to my theory is that when I have looked at the
XSLFO member-only lists and minutes at W3C (preparatory to reviewing the
XSL-FO drafts for my organization) I don't find a lot of discussion
comparing different products: instead they seem exemplory in developing
a good model and finding workable syntaxes to support it. 

In the long term, we can expect the difference between PC word
processors and browsers to  become a matter of skin rather than the
code-base. I would expect the XSL-FO model to be important because it
may provide the way to allow more satisfactory cut-and-paste into XML
applications.  

I think this is one reason why it is important that XSLT is purely
functional without side-effects: it *may* allow clearer reasoning about
transformations from model XML to the view XML to allow the inversion of
XSL transformations, so that one can edit the view and have the model
change appropriately too.   This kind of thing is not on the agenda now
AFAIK, but it is the logical next step: the difficulty of cut-and-paste
has been an enormous problem for XML systems, and has greatly impeded
product development. If FOs are well-enough analyzed that they can form
a stable basis for formatting systems, and if XSLT is anayltically
simple enough that one can reason about its effects clearly, then XML
WYSIWYG editing may become a problem tractable enough to get some nice
solutions.


So if CSS starts from utter simplicity and is attempting to grow richer,
and if XSL-FO starts from high-quality and attempts to make it as simple
as possible, what is left? Probably there is scope for a middle
language. But I think this a market that is well-catered for. If you
look at Word Perfect, FrameMaker+SGML, Cost, etc, there are many
products. HTML is getting richer; free DOCBOOK implementations exist. 
The thing missing in the middle ground, for open source, is not so much
the availability of formatters, but  the lack of GUI-based
editing/publishing environments. 

Rick Jelliffe
Academia Sinica

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************