Excluding attributes from [ID] parsing (Was: Re: [whatwg] Web Forms 2.0 comments)

[Sander]

Ian Hickson wrote:
>>> But how often would you put user-submitted data in _template_ elements?
>>
>> Basically every time I'd use a template in a CMS 'generated' by a 
>> meta-CMS.
> 
> Interesting. If you are generating templates in a CMS that will then be
> used to generate the final pages, though, what I would do would be to
<snip>

I might have misunderstood how what you propose would be applicable, but I
suspect I should have been more precise with defining what I meant by
"generated" instead (it's not a run-once generation - it's a CMS
dynamically generated based on data from a 'higher' CMS).

Ok, a use case:
A multitude of book review sites, all hosted and administrated by the same
person, but each dealing with a clearly defined genre and run from
day-to-day by independent (and essentially untrusted) reviewers.
Within each genre, short story anthologies exist. In the CMS for a given
review site a reviewer can create information on a new anthology (title,
ISBN, price, cover image) _plus_ all the stories to appear within it (a
number usually ranging from 5 to 30, but semi-regularly dipping outside
those boundaries). _Each_ story will have general information associated
with it (author, title, first publication date, review), as well as _genre
specific information_ (Example: stories within the "science fiction" genre
will have information on "awards won"; stories within the "star trek" genre
will have information on "spin-off series" as well as maybe which "main
characters" the story focuses on, etc.)
One story equals one repetition block. The initial template will seed five
"stories", with a big "add" button which reviewers will keep clicking until
they can fit in all the stories for the anthology they're reviewing.

Now if this CMS was one of a kind, there would be no problem. Unfortunately
the main administrator of these sites was lazy, so created a meta-CMS. From
within this meta-CMS, anyone who wants to create a review site for a new
genre can do so, and not only can various fields unique for the genre be
specified for short stories (which upon submission will automatically be
created in the database; and so what happens in the genre CMS is that the
fields it displays and submits are generated after they've been retrieved
from a table on "genre specific fields" in the database), but even initial
values for these fields can be specified. (The vast majority of "star trek"
short stories will be based on the "TNG" spin-off and focus on "picard and
data", so that is what's set as the initial values for those fields in the
story template.)

So that's how I get user-submitted data (the initial values; and maybe even
pattern attributes and the like for capable users who want to more strictly
limit how genre-specific fields can be used) in template elements. 
(Such meta-CMSs are not something I create every day, but still something
I've created more than once.)


Anyway, I can live with the non-breaking zero-width space character
solution, but a way to exclude all attributes of a certain type in one go
would still be very welcome.

Sander

P.S. Section 3.7.1 of the current draft still speaks of the repeat
"element".