RE: Assoc SS issue list

Personally, I think this exchange hardens my position that
the AssocSS spec should say as little as possible about
the values of the pseudo-attributes.  As long as it can
parse the PI into an attribute assignment to a known 
pseudo-attribute, it should pass that value on to the
application which can do whatever validation it wants.

paul

> -----Original Message-----
> From: John Cowan [mailto:cowan@ccil.org]
> Sent: Tuesday, 2009 June 23 23:47
> To: Simon Pieters
> Cc: John Cowan; Grosso, Paul; public-xml-core-wg@w3.org
> Subject: Re: Assoc SS issue list
> 
> Simon Pieters scripsit:
> 
> > Are you saying that what I described above is on the application
> level?
> 
> Just so.
> 
> > I think we should either have useful syntax rules, or not constrain
> the
> > syntax at all. Having arbitrary syntax rules are not useful for
> anyone,
> > IMHO.
> 
> Sure.  But consider language tags, which have three levels of
validity:
> 
> 1) A syntax consisting of one or more hyphen-separated subtags, each
of
> which consists of 1-8 alphanumeric characters (a "well-formed language
> range");
> 
> 2) The ANBF grammar in RFC 4646, soon to be replaced by a slightly
more
> restrictive grammar in RFC 4646bis (a "well-formed language tag");
> 
> 3) The ABNF grammar with the further restriction that all subtags
> appear
> in the Language Subtag Registry (a "valid language tag").
> 
> People who tag documents should of course only employ valid tags.
> However, there is usually no need for processes which accept language
> tags
> to validate them; it often suffices to check for well-formedness and
> then
> see if the tag is one of the small subset that the particular process
> has facilities to handle.  If only a fixed list of tags are meaningful
> to the process, then even well-formedness checking can be dispensed
> with.
> 
> Likewise, there is no need in my opinion for a stylesheet-pi processor
> to do more than check a pseudo-attribute value for well-formedness.
> It's up to the application, which understands the context of use, to
> determine which syntactically legal values are actually meaningful.
> The current proposal has the processor do no syntactic validation
> at all; I'm proposing something a bit more than that in the case
> of the href, type, charset, and alternate pseudo-attributes.
> 
> --
> Newbies always ask:                             John Cowan
>   "Elements or attributes?
> http://www.ccil.org/~cowan
> Which will serve me best?"                      cowan@ccil.org
>   Those who know roar like lions;
>   Wise hackers smile like tigers.                   --a tanka, or
> extended haiku

Received on Wednesday, 24 June 2009 13:12:24 UTC