315711 - Use frozen XHTML2 namespace http://www.w3.org/2006/xhtml2/

Status: RESOLVED WONTFIX

(Core :: DOM: Core & HTML, defect)

Description

[Aaron Leventhal]

The XHTML2 namespace has just been frozen by the W3C. This issue was escalated all the way up to Tim Berners Lee to get it into Firefox 1.5 in time.

We need to grab the new string for insertion into Firefox 1.5 final.

If we don't, then all future users of DHTML accessibility will have to check the Firefox browser version and dynamically change their xmlns value for xhtml2.

Needs to include closing slash:
http://www.w3.org/2006/xhtml2/

Comment 1

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Setting blocking flag so that this is seen by the right people

Comment 2

[Anne (:annevk)]

(From what I read in the HTML WG minutes it was going to be 2005. Anyway, if you heard it from TBL it must be ok :-).)

Comment 3

[Anne (:annevk)]

Also, wasn't it standard practice to implement standards that have not yet reached CR using some kind of prefix or in this case, a different namespace?

Comment 4

[Aaron Leventhal]

DHTML a11y examples using the new namespace:
http://www.mozilla.org/access/dhtml/2/

Comment 5

[Aaron Leventhal]

Attachment: Constant changes for trunk

Comment 6

[Scott MacGregor]

You should focus on getting this reviewed and landed on the trunk and we can look into it for the follow up release to 1.5. 1.5 bits are already finished It's unlikely that we would respin and slip the release for this change. Nominating for 1.8.1

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

According to the minutes (in particular, the second one):
  http://www.w3.org/2005/11/07-html-minutes#item01
  http://www.w3.org/2005/11/08-html-minutes#item06
that constant looks wrong (per comment 2).

FWIW, this is one of the reasons I made bug 290260 comment 9, but not the only one.  The semantics of those attributes could still change, even if the namespace doesn't.  (For example, how they apply to elements in other namespaces could change.)

Comment 8

[Christian :Biesinger (don't email me, ping me on IRC)]

(gecko 1.5final is long gone, adjusting target milestone)

Comment 9

[Aaron Leventhal]

Comment on attachment 202392 [details] [diff] [review]
Constant changes for trunk

Clearing out review requests until we get an explanation on the namespace discrepency from Steven Pemberton.

Comment 10

[Aaron Leventhal]

(In reply to comment #7)
> FWIW, this is one of the reasons I made bug 290260 comment 9, but not the only
> one.  The semantics of those attributes could still change, even if the
> namespace doesn't.  (For example, how they apply to elements in other
> namespaces could change.)

What is a real example of how semantics might change how the xhtml2:role attribute applies to its element?

Comment 11

[Asa Dotzler [:asa]]

It came too late for this release. We'll have to look into it for the next release.

Comment 12

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(In reply to comment #10)
> What is a real example of how semantics might change how the xhtml2:role
> attribute applies to its element?

Does it have the same meaning when on an element of any namespace, or does it apply only to XHTML2 elements?  (Does our implementation apply it to all elements (e.g., SVG, MathML, XUL, XForms), or just to XHTML1 elements?)

Comment 13

[Aaron Leventhal]

(In reply to comment #12)
> (In reply to comment #10)
> > What is a real example of how semantics might change how the xhtml2:role
> > attribute applies to its element?
> 
> Does it have the same meaning when on an element of any namespace, or does it
> apply only to XHTML2 elements?  (Does our implementation apply it to all
> elements (e.g., SVG, MathML, XUL, XForms), or just to XHTML1 elements?)

The power of it is that it applies to anything renderable. Our impl applies it to anything renderable in Mozilla -- SVG, MathML, XUL. 
- We already use it in XUL within the Firefox UI
- This is the first real step torward making accessible diagrams via SVG.
- It works with MathML as well, although I'm not sure why that would be useful.
- I haven't actually tried it with XForms yet, and I could imagine that it doesn't work yet because accessible objects are currently being created just for the anonymous HTML form control children.

Comment 14

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I cited that as something that could change, affecting the conformance of our implementation.  Your response stated the advantages of the current semantics, but didn't address their stability (especially under review from other W3C groups, which hasn't really happened yet).

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Have you gotten an explanation of the namespace discrepancy yet?  There's a chance there may be respins.

Comment 16

[Aaron Leventhal]

(In reply to comment #15)
> Have you gotten an explanation of the namespace discrepancy yet?  There's a
> chance there may be respins.

The decison to get the xhtml2 namespace change occurred with Tim Berners-Lee after meetings ended for the joint xhtml2 and Xforms face to face meeting. 

The namespace was aggreed to as:
http://www.w3.org/2006/xhtml2/

I'll post a screenshot of the IRC chat which finalized it.

Comment 17

[Aaron Leventhal]

Attachment: IRC screenshot of namespace change

Comment 18

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 202392 [details] [diff] [review]
Constant changes for trunk

I'm not likely to get to this until closer to the end of the week... (and I think I share the concerns David has about this not being a stable part of the specification).

Comment 19

[Aaron Leventhal]

(In reply to comment #18)
> (From update of attachment 202392 [details] [diff] [review] [edit])
> I'm not likely to get to this until closer to the end of the week... (and I
> think I share the concerns David has about this not being a stable part of the
> specification).
> 
Boris, are you concerned about the stability of the namespace decision?

Here is the 8:47 irc log entry for which I attached the screenshot:
http://www.w3.org/2005/11/10-xforms-irc

08:47:00 [Steven]
            FInal namespace URI for XHTML2 is: http://www.w3.org/2006/xhtml2/ 
08:48:58 [Steven]
            Team-only link to TIm's message: http://lists.w3.org/Archives/Team/w3t-archive/2005Nov/0205.html

Comment 20

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

The issue isn't that the namespace might change, but rather that the rest of the specification might change. If we ship an implementation of some xhtml2 features using the real xhtml2 namespace, we and everyone using those features will have a lot of trouble if the specification is changed.

Comment 21

[Aaron Leventhal]

(In reply to comment #20)
> The issue isn't that the namespace might change, but rather that the rest of
> the specification might change. If we ship an implementation of some xhtml2
> features using the real xhtml2 namespace, we and everyone using those features
> will have a lot of trouble if the specification is changed.
> 

No one in the WG wants to break the one implementation of this stuff. Everything is being designed to be backward compatible with Firefox. Last week they had me in Germany where I presented on the Firefox 1.5 implementation of DHTML accessibility so that stuff doesn't get broken. Everyone's on the same page.

Comment 22

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 202392 [details] [diff] [review]
Constant changes for trunk

This is ok, I suppose, but as sicking and dbaron said the problem is that the WG might _have_ to change things in the face of feedback from elsewhere.  So I'd really rather not ship this in our releases until the spec is in CR...

Comment 23

[Mike Connor [:mconnor]]

Comment on attachment 202392 [details] [diff] [review]
Constant changes for trunk

r=me subject to boris' caveats, this might not even make Firefox 2 depending on the rest of the w3c puzzle pieces

Comment 24

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Actually, i see no harm in taking this for Fx 2.0. We got the patch too late for 1.5, but for 2.0 we've got plenty of time, both for the patch and to ensure that the spec has settled

Comment 25

[Aaron Leventhal]

Instead of taking the working group's word for that backward compatibility will be maintained to the utmost, I agree it's wise to wait for the spec to settle. We can check in the xhtml2 namespace change when it's clear that we'll have backward compatibility / graceful degradation at a minimum.

Comment 26

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Doesn't appear to be critical for 1.8.1, but we'd consider a patch (especially one that was checked in to the trunk).

Comment 27

[Aaron Leventhal]

Do we officially know that this will be the XHTML2 namespace? How can we be certain?

Comment 28

[Anne (:annevk)]

You can't really be certain. At this point the draft hasn't even reached LC. It can be superceded by some other idea as well, in theory.

Comment 29

[Aaron Leventhal]

No longer an a11y issue at all. The role attribute has been added to xhtml 1.x namespace. As Anne said in comment 28 we can't fix this until we're certain about the final xhtml2 namespace.

Comment 30

[Boris Zbarsky [:bzbarsky]]

xhtml2 is dead.  RIP.