358257 - Add support for XTF attributes

Status: RESOLVED WONTFIX

(Core Graveyard :: XTF, enhancement)

Description

[Olli Pettay [:smaug][[e-mail]]]

XForms needs xtf attributes, so that bindings to non-xforms xml languages can be implemented.

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Attachment: like this

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Attachment: an example component for xtf attributes

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 243698 [details] [diff] [review]
like this

I hope this doesn't get slammed because it adds some new code. The size of XTF was just decreased in Bug 355100.

Comment 4

[aaronr]

If this gets in the trunk, it would be nice to get it on the branches, too.  At least 1.8.1.

Comment 5

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

So is the only reason for this so that we can move xmlevents into XTF? I sort of like that we support xml events and think that we should keep it.

Are there other reasons for XTF attributes? If not I think we should WONTFIX this bug.

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

There is also http://www.w3.org/TR/xforms/slice9.html#ui.repeat.via.attrs

Comment 7

[alexander :surkov (:asurkov)]

There is also http://www.w3.org/TR/xforms/slice7.html#ui-binding-foreign

Comment 8

[aaronr]

might make tables more doable in xforms.  Which currently is very difficult to get right for a form author.  But I haven't verified it, yet.

Comment 9

[Olli Pettay [:smaug][[e-mail]]]

I'd still like to have this, even if not used for XML Events.

Comment 10

[Alex Vincent [:WeirdAl]]

smaug: What happens when we have a XTF-based element accepting a XTF-based attribute?  Is it elementWrapper.willSetAttribute() and then nsIXTFAttribute.willChangeOwner, or the other way around, or is it random?  Does it follow a similar order for detaching the attribute?

Comment 11

[Olli Pettay [:smaug][[e-mail]]]

XTF elements don't handle namespaced attributes, so only the module 
implementing new XTFAttributes sees the change.

Comment 12

[alexander :surkov (:asurkov)]

Can we get any progress on this? It sounds like unique right way to get a
support of attribute based repeats for XForms.

Comment 13

[Jim Manico]

I ran into this bug today - it's blocking https://bugzilla.mozilla.org/show_bug.cgi?id=340515 - thanks for all your help.

Comment 14

[Alex Vincent [:WeirdAl]]

smaug:  I might be willing to update the patch (it has XTFElement code which is already in the tree).  I see three possible use cases:

(1) A XTF attribute node wants to implement additional interfaces.
(2) A XTF attribute node, by its presence, wants to give its ownerElement additional interfaces and properties.  This is comparable to an element whose XBL binding defines additional interfaces and properties.  For XBL, a class attribute (or other method of invoking a CSS selector and -xbl-binding rule) is how this is done.  (<implementation implements="nsIFoo"/>)
(3) A XTF attribute node wants to give its owner element additional (anonymous? native anonymous?) content.  This is what bug 340515 requests, and is comparable to XBL 1's <content/> element.

I don't see any use cases for (1).  (2) is personally appealing to me for the capabilities it might offer, but could be non-trivial.  (3) is something I think would be tougher, and would need a clearer specification.

Comment 15

[alexander :surkov (:asurkov)]

Olli, Jonas. What blocks this patch? How can we get it?

Comment 16

[Olli Pettay [:smaug][[e-mail]]]

Someone (I?) needs to update the patch.

XBL2 doesn't cover namespaced attribute, so we could perhaps take this.
(And XTF is anyway quite simple and small code.)

Comment 17

[Sergey Reymerov]

Attachment: patch for xtf attributes

i updated the patch for last version of mozilla

Comment 18

[Sergey Reymerov]

Olli, how can I test it? I made component from https://bugzilla.mozilla.org/attachment.cgi?id=243701 and made simple JS for instantiate class nsIXTFAttribute. Is it enough?

Comment 19

[Olli Pettay [:smaug][[e-mail]]]

Is that automatic? IIRC mochitests or at least chrome tests can add
new components and test them.

Comment 20

[Sergey Reymerov]

No, it's manual. Unfortunatly I can't understand how can I test it. And is it enough to create object of XTFAttribute?

Comment 21

[Sergey Reymerov]

Attachment: patch for xtf attributes

patch for mozilla (tested on 3.6 and trunc). 
also it contains simple test that creates factory for XTFAttributes (xpcshell-tests)

Comment 22

[alexander :surkov (:asurkov)]

Sergey, you should ask someone for review and superreview. Jonas will be probably the best.

Comment 23

[alexander :surkov (:asurkov)]

btw, use -p option when you make a diff.

Comment 24

[aaronr]

Comment on attachment 409948 [details] [diff] [review]
patch for xtf attributes

Olli can review this, too, he's got a lot of XTF knowledge.

Comment 25

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 409948 [details] [diff] [review]
patch for xtf attributes

Well, this is pretty much my patch updated for trunk, so I shouldn't review it.

Comment 26

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

Comment on attachment 409948 [details] [diff] [review]
patch for xtf attributes

I really don't want to do this. Attribute nodes should go away, and so should the XTF code. Attribute nodes will probably go away as soon as we can find someone to update the webcore spec, and XTF can go away once XBL2 exposes the same capabilities.

Neither of this is about to happen very soon, but I'm reluctant to spend time on adding functionality to code that is doomed.

Do feel free to rerequest review if there really is no other way to do this

Comment 27

[alexander :surkov (:asurkov)]

Olli, do you keep in mind any other way? Does it require to reorganize entirely the way how XForms works? If so then I would love to get this patch landed.

Comment 28

[Alex Vincent [:WeirdAl]]

XTF has been removed from the tree.  [xtfisdead] for filtering purposes.