996031 - crashtest layout/generic/crashtests/455407.html connects to mozilla.org everytime

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Patrick McManus [:mcmanus]]

The test infrastrucutre is meant to be self contained - so xpcshell, mochitests, and crashtests need to be self hosted.

crashtest layout/generic/crashtests/455407.html connects to mozilla.org everytime

should be fixed to be localhost only

Comment 1

[Nathan Froyd [:froydnj]]

Attachment: modify 455407.html crashtest to connect to example.org instead of mozilla.org

This patch should fix the mozilla.org issue, though I think other local changes
are interfering with testing locally, so I'm going to wait until try results
come through before asking for review.

Comment 2

[Nathan Froyd [:froydnj]]

Attachment: modify 455407.html crashtest to use a data URI instead of http://mozilla.org

Figured out what the local testing issue was.  And since reftests/crashtests
don't run proxies, we can't just connect to example.com; we need to have
something else instead.  A data URI seems appropriate.

Comment 3

[Daniel Holbert [:dholbert]]

So, this is considerably simplifying the testcase (by making the inner iframe load a blank page instead of mozilla.org).

This may be fine, but it's possible that this change prevents the testcase from triggering the original bug, rendering it useless.

Could you test the current & modified testcase against an affected build?  (Looks like a nightly from  2008-09-08 should be able to test it, base on bug 455407 comment 0. Nightlies that old may not run on recent mac/linux, but I'd expect them to run on Windows.)

Hypothetically, if the current test crashes in affected builds & your modified version does not crash, then we should not take this patch. We'd probably need a patch with a more complex data URI (more similar to mozilla.org in some relevant way), or failing that, a test-disabling patch.

Comment 4

[Daniel Holbert [:dholbert]]

(It's also possible that the current test no longer crashes in affected builds, due to changes on mozilla.org. If that's the case, then it may be better to just drop the test, since then it wouldn't actually be functioning as a unit test for its original bug anymore.)

Comment 5

[Nathan Froyd [:froydnj]]

(In reply to Daniel Holbert [:dholbert] from comment #3)
> Could you test the current & modified testcase against an affected build? 
> (Looks like a nightly from  2008-09-08 should be able to test it, base on
> bug 455407 comment 0. Nightlies that old may not run on recent mac/linux,
> but I'd expect them to run on Windows.)

I grabbed a nightly from 2008-09-07:

http://ftp.mozilla.org/pub/firefox-old-nightlies/2008/09/2008-09-07-06-mozilla1.9.0/

and ran the testcase with and without the patch.  No issues in either case.  (Though it seems poor form to have a test that just loops until timeout...)  Also tried:

http://ftp.mozilla.org/pub/firefox-old-nightlies/2008/09/2008-09-08-03-mozilla1.8/

nightly on Windows, no problems there either.  Perhaps we should just delete the test?

Comment 6

[Daniel Holbert [:dholbert]]

OK. In that case, it sounds like the testcase was unintentionally depending on some piece of www.mozilla.org to reproduce the issue, and www.mozilla.org has now changed.

Given that, there's probably no way to make the test useful again, aside from maaaybe some trial-and-error with archive.org.

So, I'd support removing it, but I'll CC mw22 [test author] in case he has any additional information / objections / suggestions.

(Also: given that bug 455407 was resolved WORKSFORME & fixed by an unknown patch, it's possible that its issue is also being tested by tests that landed with the unknown patch.)

Comment 7

[Daniel Holbert [:dholbert]]

[er, sorry - needinfo-autocomplete-fail. Meant to tap mwargers].

Comment 8

[Martijn Wargers (dead)]

I'm fine with removing the test.
Indeed, the http://www.mozilla.org reference was probably necessary for some reason to trigger the bug and apparently, I was not able to minimize the testcase without using that external reference.

Comment 9

[Nathan Froyd [:froydnj]]

Attachment: remove 455407.html crashtest

OK, removing it is much easier!

Comment 10

[Nathan Froyd [:froydnj]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/d8efaa0f46d5

Comment 11

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/d8efaa0f46d5

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/65aecee8b069
https://hg.mozilla.org/releases/mozilla-beta/rev/6431641fb1b6
https://hg.mozilla.org/releases/mozilla-b2g28_v1_3/rev/782e03676817
https://hg.mozilla.org/releases/mozilla-b2g26_v1_2/rev/19765ef79eba
https://hg.mozilla.org/releases/mozilla-esr24/rev/fa9f5d048e52