946728 - DragDrop unit tests fail

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Gaia::Homescreen, defect)

Comment 1

[Cristian Rodriguez (:crdlc)]

Attachment: Patch v1

Comment 2

[Cristian Rodriguez (:crdlc)]

I am going to disable all dragdrop tests because they are failing and I don't know why it is happening. We have to investigate the root case.

Comment 3

[Cristian Rodriguez (:crdlc)]

Disabled

https://github.com/mozilla-b2g/gaia/commit/971bea20a976d1f214ca179029d181e95f569f28

Comment 4

[Cristian Rodriguez (:crdlc)]

Hi guys, any suggestion why it could be happening? I haven't touched that code some months ago

Comment 5

[Julien Wajsberg [:julienw]]

ENSURE_DRAG_END_DELAY is 1sec, it's what causes the delay in the "end(" part of the test.

One simple change could be to use sinon's fake timers (`this.sinon.useFakeTimers()` in setup) and use `this.sinon.clock.tick(1000)` at the end of the "end" function.

The "move(" call takes 400ms and the reason is probably the same, although I don't find where it is (the onpageready event is more difficult to track ;) ). You could try to add `this.sinon.clock.tick(400)` at the end of the "move" function and see how this works ;)

You might need other calls to "this.sinon.clock.tick()" (without a value) because we have a lot of setTimeout without values, but maybe the other calls would make this work without this.

Comment 6

[Julien Wajsberg [:julienw]]

> Hi guys, any suggestion why it could be happening? I haven't touched that code some months ago

It's like this since months :p

Comment 7

[Cristian Rodriguez (:crdlc)]

but today every time jijij

Comment 8

[Julien Wajsberg [:julienw]]

note: I don't know why they fail (they don't here), but my comment 5 explains why they're slow. But this could be the cause (Travis does not like setTimeouts)

Comment 9

[Julien Wajsberg [:julienw]]

I tried to do my comment 5 but it doesn't work right now.

I'd like to try to redo the tests without loading page.js, it should be easier to debug... But is is a big work :(

Comment 10

[Cristian Rodriguez (:crdlc)]

Well, I've implemented it [1] and each test runs in 50-80ms in my pc. We will see what happens in Travis

[1] https://github.com/crdlc/gaia/commit/3d0dd7e8ac839b0fb2871e5a7b53dfb7769b26c6

Comment 11

[Julien Wajsberg [:julienw]]

Attachment: WIP patch

Does not work at all yet, but this is for reference.

Comment 12

[Julien Wajsberg [:julienw]]

(In reply to Cristian Rodriguez (:crdlc) (OFF 12/06 -- 12/10) from comment #10)
> Well, I've implemented it [1] and each test runs in 50-80ms in my pc. We
> will see what happens in Travis
> 
> [1]
> https://github.com/crdlc/gaia/commit/3d0dd7e8ac839b0fb2871e5a7b53dfb7769b26c6

It's still using timeouts, so based on past experience this will trigger intermittents for sure.

Comment 13

[Cristian Rodriguez (:crdlc)]

for sure, it was just a test. In my pc is 50ms instead of 1400ms. You can follow working on your patch and when I will come back I can be your reviewer ;)

Comment 14

[Cristian Rodriguez (:crdlc)]

Julien, please, feel free to take this bug because you created a patch to fix this. I prefer to focus on download manager bugs right now

Comment 15

[Julien Wajsberg [:julienw]]

Taking it, not sure if I'll have the time to work on this but we'll see.

Don't hesitate to steal if you want to work on this.

Comment 16

[Cristian Rodriguez (:crdlc)]

Stolen :)

Comment 17

[Cristian Rodriguez (:crdlc)]

Attachment: 16948.html

I've killed the asynchrony :) 

Before 11.05s -> After 0.82s

Comment 18

[Julien Wajsberg [:julienw]]

Comment on attachment 8386710 [details]
16948.html

r=me

thanks a lot, this was today's good surprise :)

Comment 19

[Cristian Rodriguez (:crdlc)]

:), ok going to github, thx

Comment 20

[Julien Wajsberg [:julienw]]

Will probably fix bug 938651 too

Comment 21

[Cristian Rodriguez (:crdlc)]

Merged in master:

https://github.com/mozilla-b2g/gaia/commit/f73cf17b9ad945084fdf852244fe332f261cc540