813851 - Canvas gets confused about its path/transform state

Status: RESOLVED FIXED

(Core :: Graphics: Canvas2D, defect)

Description

[Jeff Muizelaar [:jrmuizel]]

Attachment: Test case

This test case shows the problem.

Comment 1

[Jeff Muizelaar [:jrmuizel]]

Attachment: Clear out path.

I believe this is the correct fix and seems consistent with the code above, but I'm not sure.

Comment 2

[Nick Cameron [:nrc]]

Comment on attachment 683874 [details] [diff] [review]
Clear out path.

Review of attachment 683874 [details] [diff] [review]:
-----------------------------------------------------------------

I think this is right, seems consistent with the way ds path builder and mPath work elsewhere.

Comment 3

[Bas Schouten (:bas.schouten)]

Comment on attachment 683874 [details] [diff] [review]
Clear out path.

Review of attachment 683874 [details] [diff] [review]:
-----------------------------------------------------------------

Correct!

Comment 4

[Benoit Girard (:BenWa)]

Jeff is away this week so I will land this. Meanwhile we should ship this fix as early as possible. Lets track this bug.

Comment 5

[Benoit Girard (:BenWa)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/fb979e824816

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/fb979e824816

Comment 7

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Can we get the test checked in?

Comment 8

[Alex Keybl [:akeybl]]

(In reply to Benoit Girard (:BenWa) from comment #4)
> Jeff is away this week so I will land this. Meanwhile we should ship this
> fix as early as possible. Lets track this bug.

Is this a regression in FF17/18? Just want to better understand why this should track for release.

Comment 9

[Lukas Blakk [:lsblakk] use ?needinfo]

Given the lack of a case for uplift, removing tracking noms.  If you re-nom please explain what the user benefit or regression fix is here.

Comment 10

[Benoit Girard (:BenWa)]

I've been waiting on Jeff to get back from vacation to handle this. I'll let him renominate the issue.

Comment 11

[Jeff Muizelaar [:jrmuizel]]

Comment on attachment 683874 [details] [diff] [review]
Clear out path.

[Approval Request Comment]
Bug caused by (feature/regressing bug #): Azure canvas regression (happened on Windows first, then OS X and most recently Linux)
User impact if declined: In correct canvas behaviour.
Testing completed (on m-c, etc.): All of the canvas tests
Risk to taking this patch (and alternatives if risky): Seems quite low risk

Comment 12

[Jeff Muizelaar [:jrmuizel]]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #11)
> [Approval Request Comment]
> User impact if declined:

We ran into this on the upverter.com site. So it was being hit be real world sites.

Comment 13

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/7408b9bd9f10

Comment 14

[Manuela Muntean [Away]]

Attachment: Firefox 19.0 test case behavior

Comment 15

[Manuela Muntean [Away]]

Verified with Firefox 19 RC on a Mac OSX 10.7.5 machine.

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:19.0) Gecko/20100101 Firefox/19.0
Build ID: 20130215130331

The output is the one shown in the attachement from comment 14.

Comment 16

[Virgil Dicu [:virgil] [QA]]

Verified the fix for Firefox 20.0 beta 1 with the attached testcase from comment 0 on Mac OS X 10.7.5.

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:20.0) Gecko/20100101 Firefox/20.0 (20130220104816)

Comment 17

[Tracy Walker [:tracy]]

mass remove verifyme requests greater than 4 months old