539726 - [OOPP] test failure in test_plugin_position.xhtml

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Benjamin Smedberg]

test_plugin_position.xhtml fails when OOPP is turned on:

failed | p2 plugin width - got 200, expected 182
failed | p2 plugin height - got 200, expected 186

I compared the rendering with the pref flipped both ways and it appears there is a real problem here: I added debugging to the `pluginGetEdge` function in nptest_gtk2 and here is the output:

OOPP enabled:
pluginGetEdge(0)
  x/y/w/h: 12/156/200/200
pluginGetEdge(1)
  x/y/w/h: 12/156/200/200
pluginGetEdge(2)
  x/y/w/h: 12/156/200/200
pluginGetEdge(3)
  x/y/w/h: 12/156/200/200

OOPP disabled:
pluginGetEdge(0)
  x/y/w/h: 226/163/182/186
pluginGetEdge(1)
  x/y/w/h: 226/163/182/186
pluginGetEdge(2)
  x/y/w/h: 226/163/182/186
pluginGetEdge(3)
  x/y/w/h: 226/163/182/186

I'll put up screenshots in a bit.

Comment 1

[Benjamin Smedberg]

Attachment: Screenshot showing differences in layout

This screenshot shows actual differences in layout: it appears that at least the border isn't being computed correctly for the OOPP case.

Comment 2

[Benjamin Smedberg]

This appears to be Linux-specific.

Comment 3

[Karl Tomlinson (:karlt)]

I can't actually see any differences in layout.

The first SetWindow call has the correct dimensions and so does the plugin window.
The dimensions reported by getEdge seem to be GTK's default size (200).

Increasing the timeout here makes the test pass
http://hg.mozilla.org/mozilla-central/annotate/4e25e2e82747/layout/generic/test/test_plugin_position.xhtml#l75

I suspect this might be delay in XEmbed communication.

Comment 4

[Benjamin Smedberg]

http://hg.mozilla.org/mozilla-central/rev/d0c7bed861de