Closed Bug 552520 Opened 15 years ago Closed 12 years ago

[10.6] 1 pixel rendering difference is being mis-reported to Flash Player causing mouse response issue (Flash storage dialog won't go away)

Categories

(Core Graveyard :: Plug-ins, defect, P3)

1.9.2 Branch
All
macOS

Tracking

(status1.9.2 wanted, status1.9.1 unaffected)

RESOLVED WORKSFORME
Tracking Status
status1.9.2 --- wanted
status1.9.1 --- unaffected

People

(Reporter: cliss, Unassigned)

References

Details

(Keywords: regression, Whiteboard: not-ready-for-cedar)

Attachments

(4 files, 2 obsolete files)

No description provided.
Steps to reproduce: 1) Open Firefox on a 10.6.x system 2) Install Flash Player (latest version of 10 or 10.1) 3) Open youTube and start a video 4) Right click and select settings. 5) Click on the Tabs. ** if you cannot navigate that is the bug. If you can re size the window horizontally on pixel. Notice how the Video window on the left side changes position. Firefox is reporting one set of dimensions to Flash but is drawing the dialog at a different location. We check for this and they need to be the same result.
In which version was this reported ? 3.6, 3.7a2, 3.7a3pre ?
reported on 3.6 and reproduced on nightly both 3.6 and 3.7apre
I assume this is not an issue in 3.5? If so, can someone who can reproduce (e.g. has OS X 10.6.x) figure out when this regressed?
Summary: 1 pixel rendering difference is being mis-reported to Flash Player causing mouse response issue. → [10.6] 1 pixel rendering difference is being mis-reported to Flash Player causing mouse response issue.
Using Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.3a6pre) Gecko/20100616 Minefield/3.7a6pre, I do not get a context menu (See my previous bug filed - Bug 571135. The first time I tried to resize the window the flash plugin crashed. Note that I do see the dialog when I right click using Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.2.4) Gecko/20100611 Firefox/3.6.4. In both cases I am using Flash Version: 10.1.53.64 I also do notice a slight difference in how the video displays when I shift the window up horizontally using the 3.6.4 build.
some possible regression range (tested with 10.6.4, flash 10.1.53.64): works Gecko/20090721 Minefield/3.6a1pre http://hg.mozilla.org/mozilla-central/rev/f2a58ffcd00c broken Gecko/20090722 Minefield/3.6a1pre http://hg.mozilla.org/mozilla-central/rev/02f8bf10f441 http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=f2a58ffcd00c&tochange=02f8bf10f441
Perhaps the widgetry changes there?
I suspect bug 339548 indeed, there were other '1px' regressions in flash. Note that the behaviour fo those old builds is much worse than with current builds. Once those pref settings are in front of the movie, there is no way to dismiss it. In current builds (fx 3.6.4+ and Minefield) resizing the window 'helps' for that.
Summary: [10.6] 1 pixel rendering difference is being mis-reported to Flash Player causing mouse response issue. → [10.6] 1 pixel rendering difference is being mis-reported to Flash Player causing mouse response issue (Flash storage dialog won't go away)
Receiving many complaints that users cannot dismiss or interact with the Settings UI. The work around to resize the window is not readily known, which forces users to close the browser window or quit out of Firefox 3.6.
blocking1.9.2: --- → ?
blocking2.0: --- → ?
Priority: -- → P1
blocking1.9.2: ? → ---
status2.0: --- → wanted
This will likely change completely with out of process plugins on Mac. Either way we should look into this. Josh, can you have a look, or reassign appropriately?
Assignee: nobody → joshmoz
Marking this a blocker. Josh, if you disagree please re-mark appropriately.
blocking2.0: ? → betaN+
I can repro on 3.6 but not trunk, roc suggested that retained layers might have fixed this on trunk.
blocking2.0: betaN+ → ---
status2.0: wanted → ---
Hardware: x86 → All
Assignee: joshmoz → b56girard
Attached file testcase
Here is a test case from bug 556052. I suspect this only happens if the position is on a sub-pixel and that it is cause by rounding error.
Attached patch Fix translation v1 (obsolete) — Splinter Review
The translation code did not take into consideration sub pixel positioning. This causes the code to round up twice causing a 1 pixel difference.
Attachment #467927 - Flags: review?(joshmoz)
Removed change that didn't belong in the patch.
Attachment #467927 - Attachment is obsolete: true
Attachment #467929 - Flags: review?(joshmoz)
Attachment #467927 - Flags: review?(joshmoz)
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 I'm going to let roc review this if he's OK with that, I'm pretty sure he fixed this on trunk.
Attachment #467929 - Flags: review?(joshmoz) → review?(roc)
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 Seems like this could be tested pretty easily
Attachment #467929 - Flags: review?(roc) → review+
Here is a try-server build: http://ftp.mozilla.org/pub/mozilla.org/firefox/tryserver-builds/b56girard@gmail.com-9f1f8d953118/tryserver-macosx/ I got a ton of unrelated failure from the try-server. The results do not appear to be reliable. Perhaps I pushed wrong? The build itself appears to work well. I will be away this weekend. Can someone a? this patch, I am not sure who to ask.
Status: NEW → ASSIGNED
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 approvals don't go to any specific person
Attachment #467929 - Flags: approval2.0?
Attachment #467929 - Flags: approval2.0? → approval2.0+
Version: unspecified → 1.9.2 Branch
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 This needs to go on for 1.9.2, not 2.0.
Attachment #467929 - Flags: approval1.9.2.10?
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 Are we not going to get a test for this?
Attachment #467929 - Flags: approval1.9.2.11? → approval1.9.2.11+
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 Missed the 1.9.2.11 release
Attachment #467929 - Flags: approval1.9.2.11+ → approval1.9.2.11-
We're still getting a lot of reports on this on Firefox 3.6.x. I see it's been approved to commit on the v2 trunk, Will it make it into 1.9.2.12? Thanks!
Attachment #467929 - Flags: approval1.9.2.12? → approval1.9.2.12+
Thanks for approval. I'll land this tonight unless someone can get to it first.
Push: http://hg.mozilla.org/releases/mozilla-1.9.2/rev/759558e9b94d Backout: http://hg.mozilla.org/releases/mozilla-1.9.2/rev/2599ed882191 Backout due to test failure. The drawing region isn't correct with all cases with this patch applied. Test failing: modules/plugin/test/reftest/plugin-sanity.html
Keywords: checkin-needed
Whiteboard: [needs landing]
Attached image Test Result (obsolete) —
Attached image Test Reference
Attached image Test Result
Attachment #485531 - Attachment is obsolete: true
Assignee: b56girard → joshmoz
607923 duplicate to this. Can reproduce with Firefox 3.6.12 but cannot with 3.6.11.
I completely missed comment #25. Landed: http://hg.mozilla.org/releases/mozilla-1.9.2/rev/5b2446e3462d Backed out: http://hg.mozilla.org/releases/mozilla-1.9.2/rev/a47c2a2c2623 Anyway, this should land on trunk and work correctly first there before going on the branches.
(In reply to comment #31) > Anyway, this should land on trunk and work correctly first there before going > on the branches. Reed, this doesn't need to go on trunk, see comment 16 and comment 20. Only 1.9.2 needs this.
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 removing approval, we'd need a roll-up patch with a test fix in order to approve.
Attachment #467929 - Flags: approval1.9.2.13+ → approval1.9.2.13-
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 I looked over this a bit, I don't see anything wrong with the test and the rendering with this patch applied is incorrect. We'll have to come up with a correct fix.
Attachment #467929 - Flags: review-
Assignee: joshmoz → roc
Comment on attachment 467929 [details] [diff] [review] Fix translation v2 Doesn't apply to trunk, as per comment somethingorother above. Bookkeeping to remove the approval.
Attachment #467929 - Flags: approval2.0+ → approval2.0-
even though I might repeat myself in respect to bug 510846, I *can* confirm the "unable-to-navigate" issue here in 4.0b13pre @ Win7 Mozilla/5.0 (Windows NT 6.1; rv:2.0b13pre) Gecko/20110303 Firefox/4.0b13pre The flash widget (more exactly: the widget that asks you if your camera and microphone may be accessed by a site that wants to make use of this hardware) is dead as a doornail, no clicks work. Zooming the page does not work either, been there done that, no avail.
Whiteboard: not-ready-for-cedar
I posted the following to Bug 510846 as these two issues seem very similar. Error occurs in a camera snapshot app I created which requires the camera and mic settings dialog for flash to pop up. When the popup comes up "Allow ... to access your camera and microphone?" you can't click on ANYTHING. Neither Allow, nor Deny nor the "Remember" check box work. Additionally I am popping up the app in a new browser window width:580px height:480px. The swf object is set to a width:276px and height:360px and is being centered in the window using JavaScript. I am using Windows 7 Pro, FF 4.0.1, Flash 10.3.181.14, Flash 10.2 is also unresponsive. Mac OS 10.5.8 + FF 4.0.1 + Flash 10.3 are also unresponsive.
Assignee: bgirard → nobody
If I'm reading this bug correctly, this was a branch-only issue. If it is still an issue, please reopen.
Status: ASSIGNED → RESOLVED
Closed: 12 years ago
Priority: P1 → P3
Resolution: --- → WORKSFORME
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: