362910 - xml[elem]() does not call XML method

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Igor Bukanov]

Consider the following js shell session:

js> xml=<a><b/><c/></a>
<a>
  <b/>
  <c/>
</a>
js> xml['children']()
xml['children']()
<b/>
<c/>
js> xml[tmp = 'children']()
xml[tmp = 'children']()
typein:18: TypeError: xml[tmp = "children"] is not a function
js> 

Here xml[tmp = 'children']() instead of the expected call to XML method generates an error since SpiderMonkey does not call GetMethod for [] operator in the call context. xml['children']() works because that is optimized as xml.children().

Comment 1

[Igor Bukanov]

Attachment: v1

The patch adds js_GetMethod to consolidate various queries for a XML/!XML method property spread throughout the code. JSOP_CALLELEM uses the new method.

The patch also removes JSXMLObjectOps - they are useless as in all cases the code use OBJECT_IS_XML guards before casting JSObjectOps to them. As such the code can always call the function in question directly.

Comment 2

[Igor Bukanov]

Attachment: v2

In the previous patch I have added a bogus assert to GetXMLFunction.

Comment 3

[Brendan Eich [:brendan]]

Comment on attachment 367067 [details] [diff] [review]
v2

Looks good, we can always evolve JSObjectOps directly in bug 408416.

/be

Comment 4

[Brendan Eich [:brendan]]

This is a good cleanup that reduces code size; with baking time it can go into 1.9.1 for b4.

/be

Comment 5

[Robert Sayre]

Well, OK, but let's try to focus on blockers. Code cleanup is not a late in the cycle activity.

Comment 6

[Brendan Eich [:brendan]]

True, not sure if this is on the radar for another bug. It doesn't block anything. Igor?

/be

Comment 7

[Robert Sayre]

Igor mentioned another bug relating to this in IRC, I don't remember which it was.

Comment 8

[Igor Bukanov]

(In reply to comment #7)
> Igor mentioned another bug relating to this in IRC, I don't remember which it
> was.

Most of the cleanup (js_GetMethod) comes from an initial prototype for bug 369696 to implement "the proper" solution there. That does not work (I will comment in the bug), but at least instead of throwing it away completely that part ended up in this bug.

Comment 9

[Igor Bukanov]

landed to TM - http://hg.mozilla.org/tracemonkey/rev/4650f81f1853

Comment 10

[Robert Sayre]

http://hg.mozilla.org/mozilla-central/rev/4650f81f1853