RE: [all] Call for consensus on specialRequirements

Hi Michael, all,

Thanks for the text and examples, and to help this to move forward.

I have a few notes so far:


1) Two vs. One 

Those two constraints could probably be split into two data categories. Besides being constrains they have little in common. Having them separated may also help in implementation: some tools may implement only one rather than none.


2) List for forbidden Characters

Instead of the very basic list of Unicode code points, I think we could go the extra length and use a very simple regex based on Arle's proposal here:
http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Jul/0143.html
But maybe limiting it to the [] ^ - (and \uHHHH) functions. The notation would also make [] optional so even forbiddenChars="abc" would be fine (= "[abc]"

- Those are supported as far as I can tell by all regex engine without mapping.
- they would make ranges a lot easier to declare.


3) Do we want pointers in the global rule?

This would add the capability to map non-native-ITS constructs to be mapped to the data category/ies (as long as the values are the same).


4) Scope (a general remark, not just for this data category)

The text says: "The information applies to the textual content of the element, including child elements, but excluding attributes."

Why do we say (here and in many other places) "the information applies to the textual content of the element"? If the selector points to an attribute node (that could be translatable), then the information applies to the value of the attribute not to any textual content of the element.

Would there be a better way to word this text? Something like:

When the selection is an element node the information applies to the textual content of the element, including child elements, but excluding attributes. When the selection is an attribute node, the information applies to the value of the selected attribute.


Cheers
-yves

Received on Monday, 30 July 2012 12:48:45 UTC