471139 - mochitest failure in test_HTMLDocument12.html

Status: RESOLVED WORKSFORME

(Core :: DOM: Core & HTML, defect)

Description

[Justin Dolske [:Dolske]]

+++ This bug was initially created as a clone of Bug #427878 +++

I reenabled the dom-level2-html tests from bug 427878 last night; they had previously been disabled due to random failures about 8 months ago, prior to the FF3 release.

The tests passed locally for me, but they're reliably failing on the unit test boxes for all 3 platforms. Each time in the same test and with the same error:

  test_HTMLDocument12.html
  ERROR TEST-UNEXPECTED-FAIL | cookieLink | got "a=2", expected ""

So, I've disabled just this test for now.

Comment 1

[Phil Ringnalda (:philor)]

That's from http://mxr.mozilla.org/mozilla-central/source/content/base/test/test_CrossSiteXHR.html?raw=1 not cleaning up after itself - run just it, check cookies, you've got an a=2 for localhost, run test_HTMLDocument12.html and it fails.

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

I think I'd prefer to not have HTMLDocument12.html rely on that no cookies exist when entering the test. That seems really fragile and something that will create random failures once we add more tests that use cookies in the future.

Comment 3

[Serge Gautherie (:sgautherie)]

(In reply to comment #1)
> http://mxr.mozilla.org/mozilla-central/source/content/base/test/test_CrossSiteXHR.html?raw=1

Note that I'll be touching that test in bug 471122.
So, if you have any request...

(In reply to comment #2)
> I think I'd prefer to not have HTMLDocument12.html rely on that no cookies
> exist when entering the test.

I guess this test should rather cleanup (or count or something) any unwanted cookie/config before proceeding.

Comment 4

[[e-mail]]

yeah, just call nsICookieManager.removeAll() at the beginning of the test. (it'd be nice if the other test cleaned up after itself, too...)

Comment 5

[Ryan VanderMeulen [:RyanVM]]

Attachment: re-enable the test

Test appears to pass on Try these days, but given the discussion in this bug, I'm not sure if that's expected or not. It's possible that the original problem is masked/mitigated by aggressive chunking and run-by-dir these days. Jonas, are you OK with re-enabling it?
https://treeherder.mozilla.org/#/jobs?repo=try&revision=e81c3851994eec4a7cb2be29d0d9c94bc3441a1d&group_state=expanded

Comment 6

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 8737564 [details] [diff] [review]
re-enable the test

Review of attachment 8737564 [details] [diff] [review]:
-----------------------------------------------------------------

r=me if you add a comment like:

This test relies on that no cookies exist when the test starts. If this test is failing, add code to clear all cookies to the beginning of the test.

Preferably to the test itself.

Comment 7

[Pulsebot]

https://hg.mozilla.org/mozilla-central/rev/8ff351422619