Robin Berjon wrote:
On Aug 31, 2009, at 21:56 , Marcos Caceres wrote:
On Aug 31, 2009, at 8:06 PM, mozer <[email protected]> wrote:
On Mon, Aug 31, 2009 at 4:10 PM, Robin Berjon<[email protected]> wrote:
On Aug 31, 2009, at 15:58 , Marcos Caceres wrote:
As we have partitioned the P&C spec into multiple specs, the RelaxNG
schema is no longer complete (e.g., excludes updates and warp). I
suggest we
rip the relax NG schema out of the P&C spec and republish it as a
non-normative note. That way, we can update it separately if we
spawn any
new specs or if things change.
That's one option. Another option is to define it using groups, and
have the
other specs compose their additions in the groups. It's not very
hard to do.
+1 for using groups
Ok, what are these crazy groups you speak of? Please explain.
It's quite trivial, and one of the features that shows just how nice
RelaxNG is. Let's look at an example.
If you look at the top of
http://www.w3.org/Graphics/SVG/1.2/rng/Tiny-1.2/shapes.rng, you'll see:
<define name="svg.GCommon.group" combine="choice">
<choice>
<ref name="path"/>
<ref name="rect"/>
<ref name="circle"/>
<ref name="line"/>
<ref name="ellipse"/>
<ref name="polyline"/>
<ref name="polygon"/>
</choice>
</define>
That's one instance of a group that is the common content for containers
in SVG. But those containers can't just contain shapes, so what's going
on? If you look at
http://www.w3.org/Graphics/SVG/1.2/rng/Tiny-1.2/video.rng you get:
<define name="svg.GCommon.group" combine="choice">
<ref name="video"/>
</define>
The trick's in the @combine=choice — it combines the content of all the
<define>s of the same name according to its value (here, as if by
<choice>). You can look through the SVG Tiny 1.2 RNG and you'll see a
lot of that. At the end, modules are independent, and can easily be
augmented. It's simple, and it works really well.
Ha! yes, trick indeed!
It would be straightforward to have a P+C module with a group for the
choice of its children, and then for instance WARP could just add
<access> to that content.
Nah. I don't like it! Not only does it require a rocket-science degree,
but it's still means that you need to look in X-number of places to
collate all the bits of the schema. Like the Internets, this should all
be centralized ;)
Furthermore, since the schema isn't normative anyway, I think it could
just sit in CVS like a software project and be pointed to by the specs.
No need to include it in the spec's body.
Yes. This I like. Simple and centralized:
http://dev.w3.org/2006/waf/widgets-schema/widgets.rng
I've removed it from P&C.
Kind regards,
Marcos
