763470 - False positive test for cut-off elements for advanced pane tabs

Status: RESOLVED FIXED

(Mozilla QA Graveyard :: Mozmill Tests, defect, P2)

Description

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Our Mozmill l10n tests have been shown that there are some elements which are cut-off due to the localization. The results of the test you can find here:

http://mozmill-ci.blargon7.com/#/l10n/report/87961186c2b807b7747aa7e6d40d93bd

Details:

 Node is cut off by 0 px at the left: prefpane#SignonViewerDialogPane. Parent node: deck[flex=1][anonid=paneDeck]
Node is cut off by 4 px at the bottom: tab#generalTab. Parent node: tabs#tabsElement
Node is cut off by 4 px at the bottom: tab#networkTab. Parent node: tabs#tabsElement
Node is cut off by 4 px at the bottom: tab#updateTab. Parent node: tabs#tabsElement
Node is cut off by 4 px at the bottom: tab#encryptionTab. Parent node: tabs#tabsElement
Node is cut off by 4 px at the bottom: tab#ca_tab. Parent node: tabs#certMgrTabbox

Comment 1

[Francesco Lodolo [:flod]]

Henrik, I checked Mozilla/5.0 (X11; Linux x86_64; rv:16.0) Gecko/16.0 Firefox/16.0a1 on Ubuntu and I can't see anything wrong or cut in Preferences. Any help to find out what is wrong with this results?

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Attachment: screenshot

Actually I do not see a difference in the screenshot either. I think that we have to disable this specific test until all those false positives have been sorted out.

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Ok, I have reopened bug 741301 which we will use for tracking all the existent false positives. Thanks Francesco for pointing that out!

Comment 4

[Andrei Eftimie]

Attachment: Preferences/Applications -- test

Comment 5

[Andrei Eftimie]

Attachment: Preferences/Applications -- manually

I am not sure all of them are false positives.

These images are from tests/l10n/testCropped/test1.js which fails with "Node is cut off by 26 px at the bottom: richlistbox#handlersView. Parent node: vbox[flex=1][xbl:inherits=flex][class=content-box]"

The first image is automatically grabbed during the test (with the help of bug 838192).
The second image is a print screen while manually openeing the same Preference window.

Visually it is clear that the element is cut off.

====

During the test, these are its reported dimensions (of the parentBox - xul:vbox):

Box Model:
  height: 338
JS:
  scrollHeight: 366
  clientHeight: 338

Inspecting the element in DOM inspector gives the following dimensions:

Box Model:
  height: 366
JS:
  clientHeight: 366
  scrollHeight: 366

=====

We might need to wait for something to have the correct height. Although note that the scrollHeight is correct from the start. We might be able to use that, but I have a hunch that might break in ugly ways in other places.

Comment 6

[Henrik Skupin [:whimboo][⌚️UTC+1]]

(In reply to Andrei Eftimie from comment #5)
> I am not sure all of them are false positives.

In case of scrollable elements we have known failures. See bug 763461. Looks like we should fix that first so it makes it easier for us to get rid of the false positives here.

Comment 7

[Andrei Eftimie]

Dependency has been fixed, cropped tests are now successfully running in CI.