W3C home > Mailing lists > Public > www-xsl-fo@w3.org > February 2001

Re: extensions to FO

From: David Tolpin <dvd@renderx.com>
Date: Sun, 4 Feb 2001 19:55:15 +0400 (AMT)
Message-Id: <200102041555.TAA25184@mix.aua.am>
To: sebastian.rahtz@computing-services.oxford.ac.uk (Sebastian Rahtz)
Cc: www-xsl-fo@w3c.org
> how can the FO spec say anything about it? the spec tells us the
> syntax and semantics of elements in the FO namespace, no more and no
> less. It has no means of telling me that embedded MathML or svg are
> illegal, as it would break some cardinal principles of XML. The (very
> useful!!) FO DTD from RenderX is NOT part of the spec, nor can I see
> how it ever could be.
> 
> Would some XML grand wizard like to tell me I am completely wrong?

I don't see any sense in allowing non-fo elements to appear in an fo document.
FO is neither a storage nor an interchange format; it is an intermediate format
by its nature. Publishing fo documents is senseless; processing an fo is not
the time when one should deal with filtering them. 

If a processor does not support an extension, then it should not receive it
as a part of an fo document. Publishing fo's (with or without extensions)
for any other purposes than testing engines and documenting extensions is
senseless. Since the source is not an fo but a way to produce it, be it
a pair of xml/xsl files or a database and a query.

An XSL FO formatter must not be fed with non-fo elements unless it explicitely
agrees to accept them.  There is a difference between XSLT and XSL FO. The diference
is that XSLT is designed so that it exactly knows what to do with elements from
non-'xsl' namespace; that is, to put it into the document tree it builds.
Thus the behavior of an XSLT processor is always predictable in regard to 'unknown'
elements.

Oppositely, the behavior of an XSL FO processor is not so predictable. Even
with namespaces and all the staff to differentiate between similar but
different tags, there is no standard behavior for non-standard elements.
So, even if a processor agrees to accept  fotex:bookmark, for example, from
the correct namespace, it does not guarantee that it produces expected result.
Non-standard elements are not standard, imagine, their behavior is not defined.
An XSL FO processor should not accept non-standard elements it does know about.

Sincerely,
David Tolpin
Received on Sunday, 4 February 2001 10:55:13 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 3 October 2007 16:06:08 GMT