288513 - add support for DOM 3 Attr.isId and setIdAttribute*

Status: RESOLVED WONTFIX

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

Description

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

.

Comment 1

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

Why?  How?

Comment 2

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

(In reply to comment #1)
> Why?  How?

This is related to Bug 258238. If someone wants to implement something like
X+V, we should support namespaced ID attributes. Also Bug 275196 is relevant.

And don't we want to support DOM 3 Core?

The current idea is to set an ID flag to the nsDOMAttribute, 
there should be some spare bits. Of course Bug 235512 should be fixed first.

Comment 3

[Alex Vincent [:WeirdAl]]

The W3C REC for elements shows setIdAttribute, setIdAttributeNS, and setIdAttributeNode -- nothing about getIdAttribute*.  What getters are you referring to?

Comment 4

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

oops, it should be Attr.isId.

Comment 5

[Alex Vincent [:WeirdAl]]

Okay, that makes more sense.  I was about to file a bug for Attr.isId.

To support "mutation-resistant" XPath generators that don't happily live in content (aka "external parties"), they need to know if an attribute attached to an element is an ID-type attribute.

To implement this, we need interfaces for nsIDOM3Attr and nsIDOM3TypeInfo (the latter so we don't break idl).

Comment 6

[Alex Vincent [:WeirdAl]]

Attachment: Implement nsIDOM3Attr::IsId()

This patch adds nsIDOM3Attr and nsIDOM3TypeInfo from W3C DOM Level 3 Core (as nsIDOMAttr is frozen), and leaves schemaTypeInfo under NS_ERROR_NOT_IMPLEMENTED.

Tested on http://www.mozilla.org/projects/seamonkey/start/ :
document.body.attributes[0].nodeName == "class"
document.body.attributes[0].isId == false
document.body.attributes[1].nodeName == "id"
document.body.attributes[1].isId == true

Comment 7

[Johnny Stenback (:jst)]

Comment on attachment 209657 [details] [diff] [review]
Implement nsIDOM3Attr::IsId()

- In nsDOMAttribute::GetIsId(PRBool* aReturn):

+  nsAutoString name;
+  GetName(name);
+
+  *aReturn = idAtom->Equals(name);

This will make attributes in a namespace whose name happen to match the id attribute name ('id') also appear to be id's, even if they're not. The check here needs to be:

  *aReturn = mNodeInfo->Equals(idAtom, kNameSpaceID_None);

This will do the right thing, since for now we don't support having ID's in a namespace (and this is faster and less code too, e.g. no string).

r+sr=jst with that fixed.

This obviously only fixes part of this bug (the easy part), the other part requires some non-trivial getElementById() plumbing changes to get right.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Attachment: Updated IsId() patch to jst's comments.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Checked in that patch.

Comment 10

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

Comment on attachment 209796 [details] [diff] [review]
Updated IsId() patch to jst's comments.

+#include "nsIDOM3TypeInfo.idl"

why not forward-declare this, to avoid unnecessary includes and speed up compilation time (a little bit at least)?

Comment 11

[Alex Vincent [:WeirdAl]]

Comment on attachment 209796 [details] [diff] [review]
Updated IsId() patch to jst's comments.

I believe this patch to be relatively safe for Fx2/SM1.5; it simply implements nsIDOM3Attr::IsId() for attribute nodes.

The patch has been on the trunk for a couple of weeks.

Comment 12

[Anne (:annevk)]

This functionality is removed in DOM4.

Comment 13

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

Awesome