RE: FW: Taking Associating Stylesheets Second Edition to PER [was: Agenda for XML Core WG telcon of 2010 June 30]

One comment below.

> -----Original Message-----
> From: Simon Pieters [mailto:simonp@opera.com]
> Sent: Wednesday, 2010 July 07 4:07
> To: Liam R E Quin
> Cc: public-xml-core-wg@w3.org; Grosso, Paul
> Subject: Re: FW: Taking Associating Stylesheets Second Edition to PER
> [was: Agenda for XML Core WG telcon of 2010 June 30]
> 
> On Tue, 06 Jul 2010 20:08:07 +0200, Liam R E Quin <liam@w3.org> wrote:
> 
> > On Tue, 2010-07-06 at 19:41 +0200, Simon Pieters wrote:
> > [...]
> >> > I met with Tim this morning; he'd like
> >> >
> >> > Add to the document, e.g. in the Applications para
> >> > At the time of edition 1 (1999) the meaning of these p-attributes
> was
> >> > not well specified,
> >>
> >> Appendix B already states:
> >>
> >> "The first edition of this specification was admirably brief, but at
> the
> >> same time left many details unstated."
> >
> > Maybe it needs to be up higher :(
> 
> I'm fine with that.
> 
> 
> >> > and at the time of edition 2 (2010) there is low
> >> > interoperability in the values between implementations;
> >>
> >> How do we assess that there is low interoperability in the values
> >> between
> >> implementations?
> > [...]
> >
> > I don't have a way to assess interop because the spec is too vague.
> 
> I don't buy that. :-) You don't need a spec at all to assess interop,
> you just need test cases. The test cases I'm aware of are:
> 
> http://simon.html5.org/test/xml/xml-stylesheet/
> http://hixie.ch/tests/adhoc/xml/styling/pi/internal/

> http://annevankesteren.nl/test/xml/pi/i/

> 
> I don't like putting a statement in the spec saying that there's low
> interop, especially when it's not backed up by evidence. But if putting
> it
> there means that we can publish, then sure.
> 
> 
> > There were claims made in our PER director's call that different
> > implementations do different things with a missing "media"
> > pseudo-attribute, although I'm not entirely sure that really means
> > low interop, because e.g. a Web browser isn't going to do anything
> > useful with a link to a FOSI stylesheet.
> 
> As far as I'm aware, all implementations attempt to apply the style
> sheet
> if the "media" pseudo-attribute is missing. Is there some
> implementation that does something different?

Yes, Arbortext editor only uses an association for which 
the media pseudo-attribute matches something it knows about.

Which makes sense.  Why should Arbortext editor try to use
all those CSS stylesheets when it's looking for a FOSI or
Styler or XSLT stylesheet?

I still believe most people arguing about the media attribute
are only thinking of CSS and its use of media as a way to 
determine which subset of rules in a given CSS stylesheet 
should be used.  That's not (necessarily) how other applications 
interpret the media attribute, and there is nothing in the
AssocSS spec that says that the media attribute is reserved
for use only by CSS.

paul

> 
> 
> >> > We need also to contact browser vendors and see if they are
> willing
> >> > to sit round a table & get conformance/semantics pinned down in a
> >> > future edition.
> >>
> >> Hmm. I work for a browser vendor. I joined this group in the hope to
> get
> >> conformance/semantics pinned down. But the WG consensus was to not
> have
> >> any conformance requirements at all.
> >
> > Right (well, I wasn't on the WG at that time but watched from
> outside..)
> >
> > I think Tim is agreeing with you, but also accepting the WG position
> > that this is an edited rec, not a whole new version.
> 
> Ok.
> 
> Cheers,
> --
> Simon Pieters
> Opera Software

Received on Wednesday, 7 July 2010 14:04:26 UTC