618920 - Mozmill uses sleeptime=4 in run_tests for no apparent reason

Status: VERIFIED FIXED

(Testing Graveyard :: Mozmill, defect)

Description

[Jeff Hammel]

In the run_tests and run_dir methods, Mozmill uses sleeptime=4.  On my
computer, changing this to sleeptime=0 doesn't have any noticable
consequences (except, of course, less of a sleep time).  If there is
no reason to keep this, then it should be deleted.  See discussion at 
https://bugzilla.mozilla.org/show_bug.cgi?id=585106#c25 and
https://bugzilla.mozilla.org/show_bug.cgi?id=585106#c32

Comment 1

[Jeff Hammel]

Attachment: patch against 1.5.2

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Are those the 4 mysterious seconds we wait after the browser has been opened and before the tests get started?

Comment 3

[Jeff Hammel]

(In reply to comment #2)
> Are those the 4 mysterious seconds we wait after the browser has been opened
> and before the tests get started?

That would be they

Comment 4

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Once this patch lands we should run a couple of tests to make sure we don't break anything for the first tests to run. CC'ing some folks.

Comment 5

[Aaron Train [:aaronmt]]

I'm a little concerned about slower systems, but yeah lets put this patch to trial

Comment 6

[Jeff Hammel]

pushed https://github.com/mozautomation/mozmill/commit/8ca6451a3a4bf579363f4c6bab797b17686d2a32

Will keep this bug open for the case on master as well as if any problems come up from this landing

Comment 7

[Jeff Hammel]

Should we port this to master?  How long do we want to wait?

Comment 8

[Henrik Skupin [:whimboo][⌚️UTC+1]]

As far as I can see it works great. There is no long delay anymore until the tests get run. I can't see a reason why it should affect slow systems. We don't really take care of the initial start pages at the moment. Whenever we want to work with those, we will simply have to call waitForPageLoad for all of them. I would say lets get it also pushed to master and finally fixed. Thanks Jeff.

Comment 9

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Oh, and for out test-runs (non-restart and restart tests) it will save us another 15*4s = 60s! By an overall time of around 464s for Minefield it will be a time-safer of 12.7%.

Comment 10

[Jeff Hammel]

pushed to master: https://github.com/mozautomation/mozmill/commit/02a5ce6ede9d1267297894d86c98888654008d8e

Comment 11

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Haven't experienced any problems with this change so far across platforms. Marking as verified fixed.

Comment 12

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Jeff, while testing Mozmill 2 today, I have seen that we still have a strange 4s delay right after start. Hasn't it be correctly fixed for 2.0?

Comment 13

[Heather Arthur [:harth]]

(In reply to comment #12)
> Jeff, while testing Mozmill 2 today, I have seen that we still have a strange
> 4s delay right after start. Hasn't it be correctly fixed for 2.0?

That is more likely related to switching to NSPR sockets for 2.0, we have to poll for a connection, and the polling interval is a second or two. Will look into cutting it down on startup if that's the case.

Comment 14

[Jeff Hammel]

The sleeptime=4 calls have been replaced in master. I checked the source this morning.  If there is a delay, this is not the cause

Comment 15

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Ok, so this is then related to NSPR. Heather, I will file a new bug for it.

Closing this bug because it's really fixed.