Re: [CSS21] Unclear applicability to XML

On Monday, August 29, 2005, 5:02:31 PM, Ian wrote:

IH> On Mon, 29 Aug 2005, Chris Lilley wrote:
>> 
>>> Our CR exit criteria requires significantly better implementations than 
>>> were available for @font-face as far as we could tell.
>> 
>> For the whole thing, or for the subset that is useful and implemented?

IH> We were unable to find a reliably implemented useful subset.


>>> In addition, we have neither a decent test suite for @font-face nor any 
>>> volunteers to write one, which would be another problem with moving 
>>> @font-face through to REC in CSS2.1.
>> 
>> The same applied to most of CSS 2.1, which until recently had only a 
>> copy of the CSS1 tests and no CSS2 tests at all.

IH> That is clearly not true;

Actually it was, until recently, true. The first W3C CSS 2.1 test suite
was a copy of the CSS1 tests, altered for the new format. Which was not
too surprising; after all, its not as if the CSS2 test suite could have
been similarly imported.

IH> multiple volunteers existed for the parts of
IH> CSS2.1 that remain (and they have by now started submitting tests).

Having volunteers is good, but is not the same as having tests. You also
seem to have missed the 'until recently' in my comment, which was
intended to acknowledge that some tests of CSS level 2 features now
exist.

>>> The @font-face feature in CSS2 is still in CSS2, and is also covered by 
>>> a CSS3 spec for which, if I am not mistaken, you are the editor. 
>>> Therefore we do not see any difficulty in the removal of this feature.
>> 
>> There is a certain 'difficulty' in having RECs that used to point to 
>> RECs now point to unstable working drafts.

IH> I encourage you to finish the CSS3 module, then.

My point was more general than the one feature. For example, XSL 1.0 and
1.1 make normative reference to the CSS2 specification. The impact on
XSL or removing features from CSS level 2 does not seem to have been
considered.

>>> CSS2.1 is intended for applications that style structured documents, be 
>>> they based on SGML or XML or another tree-based format. However, our 
>>> primary concern is with full implementations of CSS,
>> 
>> Please define that term.

IH> Implementations that are intended to support all the features in 2.1.

That seems slightly circular but I will let it slide for now as your
response seems to be clearer later on.

>>> not implementations of subparts of CSS required by other 
>>> specifications,
>> 
>> Such as CSS Print or CSS Mobile profile?

IH> Exactly. Those specifications are not being considered when it comes to 
IH> CSS2.1 exiting CR.

Okay.

>>> as it is the complete implementations that will help us exit CR. (It 
>>> doesn't really matter if we have interoperability on one half of the 
>>> spec in one set of UAs and another half of the spec in another set of 
>>> UAs -- that would be pretty worthless for Web authors!)
>> 
>> Certainly. Although, if the two sets of UAs were disjoint then having 
>> two specifications, each of which defined the half that a set of UAs 
>> used, would be of value.

IH> Indeed. I encourage you to finish the second specification in question.

Fair point for WebFonts, although a specification that defines a set of
CSS properties used for SVG is already a Rec since 2001. Since we seem
to agree on this point, are you saying that CSS 2.1 is primarily aimed
at UAs that gain benefit from implementing the current set of features
in CSS 2.1 (ie, the (X)HTML browsers)? If so, please state so
explicitly in the spec.

>> The specific request is to
>> 
>> a) clarify whether CSS2.1 is aimed at:
>>  1) HTML 4.x user agents only
>>  2) HTML 4.x and XHTML user agents (collectively, (X)HTML only)
>>  3) A variety of user agents, including (X)HTML, SVG,
>>  4) A wide variety of XML and (X)HTML user agents including MathML, DocBook,
>>  CML etc etc

IH> None of the above. It's aimed at any markup language that can benefit from
IH> CSS as a whole;

I assume you meant CSS2.1 there?

IH>  HTML, XHTML, DocBook, and a large number of proprietary
IH> languages such as XUL and private document languages.

Its not clear that XUL is a 'document language'. I'm glad to see DocBook
on the list of markup languages in scope for CSS 2.1 styling. Please
ensure that at least one DocBook example is in the specification.

IH>  In short, CSS 2.1 is
IH> a style sheet language that allows authors and users to attach style 
IH> (e.g., fonts and spacing) to structured documents (e.g., HTML documents 
IH> and XML applications).

The XML applications part is still unclear, see below

>> b) Justify the choice of language features (@rules, properties, etc) 
>> based on the response to a)
>>
>> In other words, please ensure the aim, scope and applicability of CSS2.1 
>> to XML is very clear. It is currently unclear.

IH> The abstract currently says "CSS 2.1 is a style sheet language that allows
IH> authors and users to attach style (e.g., fonts and spacing) to structured
IH> documents (e.g., HTML documents and XML applications)", is this not clear
IH> enough? It seems clear to me from this that images (SVG) and chemical data
IH> (CML) are out of scope.

Since the SVG images (in XML) and CML Chemical data (in XML) are both
XML applications, its not clear at all. If the intent is to apply CSS2.1
to 'document like' (eg XHTML, SpecProd, DocBook, and such like technical
documentation formats) then please say so explicitly.

If the intent is that CML, MusicML, BRML, CDI, MathML, SVG etc are out of
scope for CSS 2.1 then please state so explicitly.



-- 
 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 W3C Graphics Activity Lead

Received on Monday, 29 August 2005 15:31:36 UTC