Intermittent B2G TEST-UNEXPECTED-PASS | flexbox-inlinecontent-horiz-3c.xhtml | image comparison (==)

RESOLVED FIXED in Firefox 32



4 years ago
4 years ago


(Reporter: emorley, Assigned: ahal)



Gonk (Firefox OS)

Firefox Tracking Flags

(firefox30 unaffected, firefox31 unaffected, firefox32 fixed, firefox-esr24 unaffected)



(1 attachment, 1 obsolete attachment)



4 years ago
b2g_emulator_vm b2g-inbound opt test reftest-8 on 2014-05-28 20:47:28 PDT for push 437a98f062f0

slave: tst-linux64-spot-1047

21:11:25     INFO -  REFTEST TEST-UNEXPECTED-PASS | | image comparison (==)
Comment hidden (Treeherder Robot)
Created attachment 8430843 [details]
the "passing" reftest snapshot

This "pass" is because both testcase & reference case look like this attached image, which is just the words:
  "image not available"
which I assume comes from here (the only instance of that string in m-c, at least):

Comment 3

4 years ago
Note bug 981477 recently changed the manifest, marking this and others as fails-if, but it seems that the try run used to work out what would fail when switching to running B2G reftests oop has found a few that are more in the random-if camp.
Semi-disregard comment 2 -- it looks like the "image not available" is actually just coming from reftest-analyzer (that's where I saved the image from), and reftest-analyzer is showing that because the data URI in the log is broken.

So, we don't actually know what the image looks like. But per comment 3, if this "fails" annotation is very recent, then it presumably wants to be "random", I guess (?)
Comment hidden (Treeherder Robot)
Comment hidden (Treeherder Robot)
Comment hidden (Treeherder Robot)
Comment hidden (Treeherder Robot)
Comment hidden (Treeherder Robot)
Comment hidden (Treeherder Robot)
Just copying the data: URL out of the log works for me for looking at the image.
Flags: needinfo?(dholbert)
Ah -- copying the first one does. Copying the second one does not.  (I arbitrarily picked the second one.)  So, we do know what the image looks like after all.  Looks like this is indeed just a normal unexpected pass, then, and may just want to be random-if instead of fails-if, per comment 3.

(edmorley suggested this morning that the data-URI-truncation is from logcat (or some logcat processing tool) imposing a max-length on lines. Looks like the working data URI is 7706 characters long; the broken one is truncated to 863 characters.)
Flags: needinfo?(dholbert)
Attachment #8430843 - Attachment is obsolete: true
(to be clear: by "first one" & "second one", I'm talking about the two links below "Summary" for the test-failure, in the log at )

Comment 14

4 years ago
Created attachment 8431552 [details] [diff] [review]
change from fails to random

Ok, so seems like everyone is agreeing to change this from fails to random?
Attachment #8431552 - Flags: review?(dholbert)
Comment on attachment 8431552 [details] [diff] [review]
change from fails to random

Thanks. Sounds good to me.

(I'll bet the other "fails-if(B2G&&browserIsRemote)" tests here (and maybe all such tests?) really want the same treatment, but I suppose we can wait and see on the others.)
Attachment #8431552 - Flags: review?(dholbert) → review+
Assignee: nobody → ahalberstadt
Last Resolved: 4 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla32
status-firefox30: --- → unaffected
status-firefox31: --- → unaffected
status-firefox32: --- → fixed
status-firefox-esr24: --- → unaffected
Whoops, meant that for bug 981477.
You need to log in before you can comment on or make changes to this bug.