Open Bug 481092 Opened 15 years ago Updated 2 years ago

Determing if a value of browser.gesture.pinch.threshold can be found that reduces accidental zoom operations

Categories

(Firefox :: General, defect)

All
macOS
defect

Tracking

()

People

(Reporter: faaborg, Unassigned)

References

Details

(Keywords: uiwanted)

Some feedback I received in email awhile ago, no idea if it is still an issue
(don't have a new macbook to test get a feel for the changes):

>Hi Alex -- not sure who to send this to, so trying you ---
>
>I've just been watching my wife use FF 3.1b2, and she's accidentally pinch-
>zooming on her new macbook over and over again, making gmail almost unusable.  
>it's really getting in the way, and it's totally mysterious if you don't know 
>what's going on.

>Clearly part of this is an overally OS X problem, but it also feels that >safari's 
>model of one-pinch->one-zooming delta is less problematic for web content than 
>firefox's one-pinch -> more or less zoom depending on the extent of the pinch.  
>It seems to allow for feature discoverability with a fairly innocuous >consequence 
>in the event of occasional triggering, while Firefox can end up with people in a 
>jam quickly.

>I'm not using my regular laptop, so I don't know if you've already addressed >this 
>in nightlies.  If so, ignore!
I don't have a new macbook to test on, so cc'ing a few people at Mozilla to see what they think of the current threshold settings.
I think it's improved since it first landed. I don't find myself accidentally zooming as often as I originally did, but it does still happen.
Have you looked at bug 462935 which landed after beta 2?
I'm always using the latest nightly build on OS X 10.5 and I run into this issue several times within a day. It's really annoying when the zoom level is decreased most of the times I wanna delete an email via the drop down in Gmail. For that reason I switched it of some days ago. I'll try to play around with the default threshold. Alex, if you are interested we can meet up and I can show it. You could even test on your own.

I believe that a lot of people will be afraid of this default setting.
Flags: blocking-firefox3.1?
Hardware: x86 → All
Safari and Mail.app also honour the pinch-to-zoom setting. Do we have a sense of what their tolerance is set at?

My impression is that Firefox is a little lazier at interpreting the gesture. I find that I accidentally trigger it most frequently when I'm switching between tabs, or have just clicked somewhere ... while Firefox is handling that click, it then misinterprets my normal finger-resting-on-the-keyboard and next mouse move to be a drag. I definitely notice it more when I have more tabs open, and a higher CPU load.
I just tried pinching in Safari 4 and it seems to have reverted the "zoom only once" behavior. You can now zoom a ton with a single motion like we did for beta 2.

For those accidentally triggering the zoom, do you recall if you're pinching vertically or horizontally?

Tom: I believe you mentioned that once we think it's a particular gesture, we stick with it. So a pinch can't become a twist or just a 2-finger scroll? (And same thing the other way -- a 2 finger scroll won't pinch.)
Given by the comment of beltzner I ran a test in Safari too and noticed that I have the identical problems as when using Firefox. So it shouldn't be related to the threshold value? I tried to reproduce it with Firefox afterward but wasn't able to do so for the first moment. But when I closely was watching my thumb while doing those steps (to delete a message in Gmail), I've seen why this zooming happens for me.

My thumb is constantly moving upward so it is located above 1/3 of the height of the trackpad after a while. That means outside of the button area you had with former MacBook models. Having one big area without a sensible intersection between those areas will probably let you do this automatically. But when leaving the former button area any sliding your forefinger will cause a pinching action.

For the above action I'm probably to lazy to raise my thumb. After clicking the dropdown it still lays on the trackpad while the forefinger moves the cursor to the delete menuitem. And staying with the thumb outside of this ~1,5cm area at the bottom, one zoom out command is called.
(In reply to comment #6)
> Tom: I believe you mentioned that once we think it's a particular gesture, we
> stick with it. So a pinch can't become a twist or just a 2-finger scroll? (And
> same thing the other way -- a 2 finger scroll won't pinch.)

Is that us or OSX? I thought we were just reading the events given to us by the OS here. If it's us, we should change that behaviour, I think, as it's probably what's causing the somewhat confusing pinch interpretations.
(In reply to comment #6)
> 
> Tom: I believe you mentioned that once we think it's a particular gesture, we
> stick with it. So a pinch can't become a twist or just a 2-finger scroll? (And
> same thing the other way -- a 2 finger scroll won't pinch.)

The exclusion only applies between magnify and rotate multi-touch gestures. AppKit brackets each multi-touch sequence with calls to the beginGestureWithEvent:  and endGestureWithEvent: methods. During my initial testing, I discovered that, for some reason, AppKit would occasionally insert a call to magnifyWithEvent: in a sequence of rotateWithEvent: calls and vice versa during a single multi-touch sequence.  It seemed spurious since there might be 100+ calls to one method and one call to the other during a single multi-touch sequence.  Since I didn't want to have the user get confused by us  converting what is really one multi-touch sequence into three, I coded in the exclusion.

The scrolling gesture is handled by different code. The problem is that if the user spreads their fingers apart in a certain way AppKit will see that as the start of a magnify gesture. I've encountered this before as well.  The solution is to set the threshold correctly.

Mac OS X 10.6 (Snow Leopard) apparently will have a documented multi-touch API in AppKit. We will at least hopefully know what the magnify units are! (I'm guessing trackpad coordinates but who knows ...)
I don't think this blocks, but it certainly does warrant further investigation. Morphing slightly to better reflect what the actual goal is.

Also note that there is a configurable preference for the gesture tolerance: browser.gesture.pinch.threshold

Also also note that people reporting too many accidental zooms with 3.1b2 may find that the behaviour is improved in 3.1b3, so we should probably wait until that comes out.

Also also also note that comment #9 suggests this could be something we're getting from OSX, so do also ask yourself: am I getting accidental zooms in Firefox more than in other apps that support zooms?

Leaving uiwanted for now.
Flags: blocking-firefox3.1? → blocking-firefox3.1-
Keywords: uiwanted
Summary: Decide on the default value of browser.gesture.pinch.threshold → Determing if a value of browser.gesture.pinch.threshold can be found that reduces accidental zoom operations
(In reply to comment #9)
> Mac OS X 10.6 (Snow Leopard) apparently will have a documented multi-touch API
> in AppKit. We will at least hopefully know what the magnify units are! (I'm
> guessing trackpad coordinates but who knows ...)

Marcia, could you help out here? We have a 10.6 machine in the lab so we could give Tom the necessary information.
On Mac OS 10.7, it seems the problem about accidental zoom does not exist.
Trying to figure out the problem from UX perspective. Could you please clarify which gesture triggered the accidental zooms? 

I guess the performance must have been improved in the past yrs. If the issue does not exist on any Mac OS any more, this bug should be clear.

(In reply to Mike Beltzner [:beltzner] from comment #10)
> I don't think this blocks, but it certainly does warrant further
> investigation. Morphing slightly to better reflect what the actual goal is.
> 
> Also note that there is a configurable preference for the gesture tolerance:
> browser.gesture.pinch.threshold
> 
> Also also note that people reporting too many accidental zooms with 3.1b2
> may find that the behaviour is improved in 3.1b3, so we should probably wait
> until that comes out.
> 
> Also also also note that comment #9 suggests this could be something we're
> getting from OSX, so do also ask yourself: am I getting accidental zooms in
> Firefox more than in other apps that support zooms?
> 
> Leaving uiwanted for now.
Yuan: the pinch zoom gesture was disabled in bug 613909 for Firefox 4.

There is bug 688990 about re-implementing it in a Safari-like way and an older bug 632733 asking for a preference to enable the old implementation.
Depends on: 613909
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.