363032 - ###!!! ASSERTION: invalid arg: 'aIndex >= 0', file m:/trunk/mozilla/layout/generic/nsFrameList.cpp, line 300

Status: RESOLVED FIXED

(Core :: XUL, defect)

Description

[Alex Vincent [:WeirdAl]]

If a XUL deck's selectedIndex is set to -1, Gecko triggers this assertion.

Stack trace coming up.  This should be pretty easy to fix.

Comment 1

[Alex Vincent [:WeirdAl]]

Attachment: stack trace

The JavaScript portion of this stack trace shows a binding setting a deck's selectedIndex to -1.  Abusing the deck, maybe, but given how selectedIndex typically works, not entirely unreasonable.

Comment 2

[Alex Vincent [:WeirdAl]]

Attachment: patch, v1

Comment 3

[Boris Zbarsky [:bzbarsky]]

I'm swamped.  Please ask someone else for review on this.  One of the XUL peers (Neils) might be a good idea.

Comment 4

[Alex Vincent [:WeirdAl]]

Comment on attachment 247779 [details] [diff] [review]
patch, v1

trivial patch

Comment 5

[[e-mail]]

Comment on attachment 247779 [details] [diff] [review]
patch, v1

Two Neils are better than one ;-)

One idea I had was that GetSelectedBox should use mIndex, while IndexChanged should set mIndex in between HideBox and ShowBox and call GetSelectedBox instead of GetBoxAt.

The other idea I had was that GetSelectedIndex should return 0 or perhaps MAXINT for negative indices.

Comment 6

[Neil Deakin]

Comment on attachment 247779 [details] [diff] [review]
patch, v1

Look's OK. Neil's first suggestion seems good too.

Comment 7

[Alex Vincent [:WeirdAl]]

(In reply to comment #5)
> The other idea I had was that GetSelectedIndex should return 0 or perhaps
> MAXINT for negative indices.

Wouldn't that have the result of meaning a deck's selectedIndex becomes 0 or a very large number, instead of -1, when the app author wants to make sure nothing is selected?  Traditionally, -1 means no selection at all, on both XUL (tree.currentIndex, menulist.selectedIndex, and the aforementioned deck) and HTML (select.selectedIndex).

Comment 8

[[e-mail]]

Comment on attachment 247779 [details] [diff] [review]
patch, v1

Please file a followup bug, patch or both ;-)

Comment 9

[Robert Sayre]

Checking in nsDeckFrame.cpp;
/cvsroot/mozilla/layout/xul/base/src/nsDeckFrame.cpp,v  <--  nsDeckFrame.cpp
new revision: 1.78; previous revision: 1.77
done

Comment 10

[Alex Vincent [:WeirdAl]]

per comment 8, filed bug 363112.