Re: AssocSS issue #5 [was: Minutes for XML Core WG telcon of 2009 July 1]

On Thu, 16 Jul 2009 16:30:45 +0200, Grosso, Paul <pgrosso@ptc.com> wrote:

>> > If they are ignored, then the processor isn't even going to
>> > notice that they are duplicates.
>>
>> Hmm. This does not match what I had in mind.
>>
>> What I had in mind was more like the following:
>
> Of course, you're describing one possible implementation, which is
> not the way we'd want to state it in the spec, but I understand your
> description as a use case, and that's reasonable.
>
>>
>> 1. Parse the PI to obtain all pseudo-attributes. if the PI doesn't match
>> the allowed syntax, ignore the whole PI.
>
> I'm okay with that so far.
>
>> The syntax bans any duplicate pseudo-attrs.
>
> No, syntax can't do that.

It's not clear to me why not, but I guess it doesn't really matter.


> Banning duplicate attributes is part
> of the semantic processing. In the case of XML, it's a WFC
> (WFC Unique Att Spec) which gets applied after the syntactic parsing.
>
> There is no a priori reason we need to ban duplicate pseudo-atts
> before ignoring unknown ones.
>
> On the other hand, there is no a priori reason we couldn't.
> It's a decision we need to make.
>
> I figured, since we were going to ignore them anyway, there
> was no reason to toss the whole PI for something we're going
> to ignore.
>
> At this point, we need to have a WG discussion about this,
> but in the interest of making progress, I suggest we consider
> the current suggested resolution as the current status quo
> which means we'd need to have a preponderance of voices (or
> some very loud ones) to change the proposed resolution here.

Ok. I see these reasons to change from the status quo:

1. For consistency with XML.
2. For easier reuse by e.g. XBL2.
3. An implementation that supports both xml-stylesheet and XBL2 would  
likely want to use the same code for parsing both PIs.

I guess I'm suggesting that we separate the parsing of the PI with the  
rules specific to xml-stylesheet PIs. This parser should ban duplicate  
pseudo-attributes (IMHO), but should not have to have knowledge about  
which pseudo-attributes are known.


> paul
>
>> 2. Pass all pseudo-attributes on for further processing.
>> 3. In this further processing, unknown pseudo-attributes must be  
>> ignored.
>> requirements about alternate="" also go here.

-- 
Simon Pieters
Opera Software

Received on Thursday, 16 July 2009 15:31:24 UTC