971691 - "day view header copy should not overflow" marionette js test is failing

Status: RESOLVED FIXED

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

Description

[Anthony Ricaud (:rik)]

Error: content (277px) is wider than container (270px)

First red travis: https://travis-ci.org/mozilla-b2g/gaia/jobs/18698014
Last green: https://travis-ci.org/mozilla-b2g/gaia/jobs/18691959

Between those two travis runs, the Gaia change is not at fault. We're also using the same B2G desktop so it's not a Gecko change either.

I'm thinking it could be a change in marionette js or npm.

Comment 1

[Miller Medeiros [:millermedeiros]]

going to take a screenshot and analyze it. but my bet (without further investigation) is that the copy "Feb 12, Wednesday" is too wide on Travis (because system font is different). - see: https://groups.google.com/forum/#!searchin/mozilla.dev.gaia/font/mozilla.dev.gaia/DrQzv7qexw4/kZUxlbeSHEQJ

Comment 2

[Julien Wajsberg [:julienw]]

Yeah, one bet is that Travis updated their system font...

Comment 3

[Julien Wajsberg [:julienw]]

Also, it does not reproduce on TBPL.

Comment 4

[Miller Medeiros [:millermedeiros]]

Attachment: travis screenshot

confirmed, the problem is on the font size/family.. screenshot from build https://travis-ci.org/mozilla-b2g/gaia/builds/18732858


my proposed fix for the problem is to only detect the amount of chars (should be smaller than 21). It doesn't provide the same confidence as checking the overflow (since as we saw in this case fonts really matter), but will be enough to catch localization problems.

expect a patch in the next 10min.

Comment 5

[Gareth Aye [:gaye] (back from PTO)]

/me ready to r+

Comment 6

[Julien Wajsberg [:julienw]]

I wish we knew what regressed though.

Comment 7

[Miller Medeiros [:millermedeiros]]

Attachment: Link to Github pull-request: https://github.com/mozilla-b2g/gaia/pull/16214/files

Comment 8

[Gareth Aye [:gaye] (back from PTO)]

Correct me if I'm wrong but I think the regression was actually caused by the day changing since Wednesday is the longest weekday string?

Comment 9

[Miller Medeiros [:millermedeiros]]

Comment on attachment 8374818 [details]
Link to Github pull-request: https://github.com/mozilla-b2g/gaia/pull/16214/files

just need to wait travis build to finish before we merge it: https://travis-ci.org/mozilla-b2g/gaia/builds/18733968

Comment 10

[Julien Wajsberg [:julienw]]

Oh Gareth, you're so smart :)

Comment 11

[Gareth Aye [:gaye] (back from PTO)]

Thank you Julien and Rik for being on top of this!

Comment 12

[Phil Ringnalda (:philor)]

(In reply to Julien Wajsberg [:julienw] from comment #3)
> Also, it does not reproduce on TBPL.

Every single run which did not fail for some other reason before it got to this test failed this way, I don't see how that's not reproducing. gaia-integration is on tbpl as "Gi", and is hidden because it mostly fails; are you maybe looking at some other suite?

Comment 13

[Miller Medeiros [:millermedeiros]]

my patch fixed the problem but the build failed because of another test "Places tests Search for previously visited URL:" (timeout exceeded) at "/home/travis/build/mozilla-b2g/gaia/apps/search/test/marionette/places_search_test.js:37:12" - I created a bug report for it: https://bugzilla.mozilla.org/show_bug.cgi?id=971725

I think we should land it anyway since this is blocking other devs. maybe later today someone can disable the "places_search_test" and create another issue asking someone else to fix it and re-enable the test.

can someone merge it for me?

Comment 14

[Julien Wajsberg [:julienw]]

(In reply to Phil Ringnalda (:philor) from comment #12)
> (In reply to Julien Wajsberg [:julienw] from comment #3)
> > Also, it does not reproduce on TBPL.
> 
> Every single run which did not fail for some other reason before it got to
> this test failed this way, I don't see how that's not reproducing.
> gaia-integration is on tbpl as "Gi", and is hidden because it mostly fails;
> are you maybe looking at some other suite?

I was looking at exactly this, using showall=1, but I think the failing builds were not finished yet. For example, see that there is a build where we have both a green and a red on 2 different configs ;)

Comment 15

[Anthony Ricaud (:rik)]

Thanks for the very quick fix Miller!

https://github.com/mozilla-b2g/gaia/commit/2a435446f729f316d36eccf8055d92a0b3764e7e

Sadly, we have unit test failures now so I can't reopen the tree yet :(