Closed Bug 1158016 Opened 9 years ago Closed 9 years ago

Intermittent test_bug386782.html | Edited contents still correct? - got <p>designModeDocument</p>, expected <p>EDITED designModeDocument</p> and more

Categories

(Core :: DOM: Navigation, defect)

ARM
Android
defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla41
Tracking Status
firefox39 --- fixed
firefox40 --- fixed
firefox41 --- fixed
firefox-esr31 --- unaffected
firefox-esr38 --- unaffected

People

(Reporter: philor, Assigned: ehsan.akhgari)

Details

(Keywords: intermittent-failure)

Attachments

(1 file)

      No description provided.
:ehsan -- Any ideas?

I note the friendly warning:

      // WARNING: If the following test fails, give the setTimeout() in the onload()
      // a bit longer; the doc hasn't had enough time to setup its editor.

but I think you removed that weakness in bug 493381.
Flags: needinfo?(ehsan)
(In reply to Geoff Brown [:gbrown] from comment #65)
> :ehsan -- Any ideas?
> 
> I note the friendly warning:
> 
>       // WARNING: If the following test fails, give the setTimeout() in the
> onload()
>       // a bit longer; the doc hasn't had enough time to setup its editor.
> 
> but I think you removed that weakness in bug 493381.

I think that comment is not important here.  execCommand is throwing NS_ERROR_FAILURE, which I bet is coming from here: <https://dxr.mozilla.org/mozilla-central/source/dom/html/nsHTMLDocument.cpp#3256>

This is also consistent with the test failure message: "Edited contents still correct? - got <p>designModeDocument</p>, expected <p>EDITED designModeDocument</p>".

All of this points to me that the page was not put in the bf-cache.  That however I can't explain.  Especially since this is only happening on Android.  :(

Boris, can you think of anything here?
Component: Testing → Document Navigation
Flags: needinfo?(ehsan) → needinfo?(bzbarsky)
Product: Firefox for Android → Core
It's entirely possible things are not ending up in bfcache; e.g. if memory pressure fires at the wrong time or something.

We could have the page that's the target of the navigation notify us when it gets pageshow, and if the bfcache bit is important it could tell us whether it came from bfcache or not, since the pageshow event has that data.
Flags: needinfo?(bzbarsky)
Note to self: check for the pageshow event's persisted attribute on the page loaded in the inner iframe to see if we're coming from the bfcache.
Flags: needinfo?(ehsan)
I have a patch which seems to fix this: https://treeherder.mozilla.org/#/jobs?repo=try&revision=1e36026245ee

Note that it seems like there are tons of other intermittent failures on mochitest-1 on Android 4.3.
Assignee: nobody → ehsan
Flags: needinfo?(ehsan)
Attachment #8611280 - Flags: review?(bzbarsky)
(In reply to :Ehsan Akhgari (not reading bugmail, needinfo? me!) from comment #86)
> I have a patch which seems to fix this:
> https://treeherder.mozilla.org/#/jobs?repo=try&revision=1e36026245ee
> 
> Note that it seems like there are tons of other intermittent failures on
> mochitest-1 on Android 4.3.

I have a bad feeling about those try results. Most of those mochitest-1 failures look new to me.
Stuff as many of the dom/animation/css-transform ones as will fit into the summary of bug 1162208, and then how would it look?

I'd expect you would find all or almost all of the remaining ones, or a nearby test doing the same thing, in the last 100 runs of it on m-i, starred as "intermittent" with no bug if someone else did it, starred as "expected fail" with the comment "a" if I did it.
I see what you mean -- thanks for the expert starring.

I may have just been luckier here (without this patch, and a different baseline) https://treeherder.mozilla.org/#/jobs?repo=try&revision=658cba3b5d1c
Comment on attachment 8611280 [details] [diff] [review]
Skip the test if the page is not loaded from the bf-cache when going back

r=me
Attachment #8611280 - Flags: review?(bzbarsky) → review+
https://hg.mozilla.org/mozilla-central/rev/9895a4e1cfb0
Status: NEW → RESOLVED
Closed: 9 years ago
Flags: in-testsuite+
Resolution: --- → FIXED
Target Milestone: --- → mozilla41
You need to log in before you can comment on or make changes to this bug.