1559798 - Begin polling for the results (runner.js) only after the pageload.js code has been invoked in the test page content

Status: RESOLVED FIXED

(Testing :: Raptor, defect, P1)

Comment 1

[Robert Wood [:rwood]]

While running Raptor locally with extra logging in the web extension I noticed that the Raptor runner.js has already started polling for the page load results - BEFORE the Raptor pageload.js content was actually invoked, i.e.:

hecking results... runner.js:265:15
setting timeout to checkForResult again in 250ms runner.js:285:21
checking results... runner.js:265:15
setting timeout to checkForResult again in 250ms runner.js:285:21
checking results... runner.js:265:15
setting timeout to checkForResult again in 250ms runner.js:285:21
pagelaodjs adding load event listener for raptorContentHandler pageload.js:333:11
checking results... runner.js:265:15
setting timeout to checkForResult again in 250ms runner.js:285:21
checking results... runner.js:265:15

Change Raptor runner.js so it starts the page-timeout timer and polling for results only after the pageload.js content has been injected and invoked. This should reduce the number of Raptor intermittent page-timeout failures in the tp6 suites.

Comment 2

[Robert Wood [:rwood]]

I put together a patch to try it out:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=3694563ab60b6a81352217c9b32af2f1f6dbb7f3

Andrew also put together a patch trying it out (I like his solution better):

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c6d0c16172656648eda09606f48b9e986b28d59e

Both show promising results so far.

Comment 3

[Robert Wood [:rwood]]

:acreskey would you like to submit your patch?

Comment 4

[Andrew Creskey [:acreskey]]

Rob, I couldn't in good faith submit my patch because it's just an iteration on your patch.
I'm happy to review what you put up though.

Comment 5

[Robert Wood [:rwood]]

(In reply to Andrew Creskey from comment #4)

Rob, I couldn't in good faith submit my patch because it's just an iteration on your patch.
I'm happy to review what you put up though.

Thanks Andrew, ok I'll put up a patch today, excited to get this change in! :)

Comment 6

[Robert Wood [:rwood]]

Attachment: Bug 1559798 - Begin polling for the results (runner.js) only after the pageload.js code has been invoked in the test page content

Comment 7

[Robert Wood [:rwood]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=d93f9a870b46ab8faa04539d53cdb1e98285a57f

Comment 8

[Robert Wood [:rwood]]

Just fixing my int issue:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=fb2504d00519df40b1b93378a0b086cdcb2a4379

Still need to update according to review comments in phab.

Comment 9

[Robert Wood [:rwood]]

Updated according to phab review comments (thanks!):

https://treeherder.mozilla.org/#/jobs?repo=try&revision=becfde72536331b76fee7d8526f7b2d1425ca772

Comment 10

[Robert Wood [:rwood]]

After a few more improvements:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=f48b41d1fc0aac87ee800ccd992cf23de6abcb73

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/9f0bf9849476
Begin polling for the results (runner.js) only after the pageload.js code has been invoked in the test page content r=acreskey,tarek

Comment 12

[Arthur Iakab [arthur_iakab]]

https://hg.mozilla.org/mozilla-central/rev/9f0bf9849476