694906 - overlay insert behavior has changed

Status: RESOLVED INVALID

(Firefox for Android Graveyard :: Extension Compatibility, defect)

Description

[Eric Coleman]

User Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.23) Gecko/20110920 Firefox/3.6.23
Build ID: 20110920075126

Steps to reproduce:

When creating an overlay, the insertion point behavior has changed.  In Fennec 7, the behavior matches Firefox.  In Fennec 8 and up the nightly as of now, the behavior has changed.  All that's been done is following the rules for creating an extension as per both Firefox and Fennec, (at least up to 7).


Actual results:

Inserting an overly as a child of a element with no ID attribute ends up making the overlay a sibling instead.


Expected results:

Here is a more detailed account of behavior changes:

Consider some chrome that looks like the following:

<hbox id="myparent">
 <vbox>
    <box id="child1" />
    <box id="child2" />
  </vbox>
<hbox>

In Firefox, and in Fennec 7 and lower, in order to put an overly as a sibling of "child1" the following code would have worked:

<overlay>
<hbox id="myparent">
 <vbox>
    <box id="myextension" insertbefore="child1">
     <!-- my extension chrome... -->
    </box>
</vbox>
</hbox>
</overlay>

That's supposed to result in the following:
<hbox id="myparent">
 <vbox>
    <box id="myextension" />
    <box id="child1" />
    <box id="child2" />
  </vbox>
<hbox>

However, in Fennec 8 and the current nightly, the following gets generated instead:
<hbox id="myparent">
 <vbox>
    <box id="child1" />
    <box id="child2" />
  </vbox>
 <vbox>
    <box id="myextension" />
 </vbox>
<hbox>

Comment 1

[Matt Brubeck (:mbrubeck)]

Nominating for tracking-firefox8 because this is a regression that affects add-on compatibility in for Fennec update 8.

Eric, do you have an example extension (either a real extension or a minimized test case) that demonstrates this behavior?  What particular element(s) are you adding children to?

Comment 2

[Eric Coleman]

Oops! I made a big mistake in filing this bug report.  My apologies to all. :)

I'm not entirely sure how I was able to observe what I did in Fennec 7.  However, the behavior in Fennec 8 and the nightly is the current behavior in Fennec 7, so there is no regression.

Comment 3

[Eric Coleman]

Can the status of this be marked as invalid?  Thanks!

Comment 4

[Matt Brubeck (:mbrubeck)]

Feel free to open a new bug if you believe the Fennec behavior is incorrect.  A test case would be great.