420967 - file picker dialog from file upload behaves strangely

Status: RESOLVED FIXED

(Core :: Widget: Cocoa, defect, P2)

Description

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Couple of problems, all seen from bugzilla attachment upload page:

- With the file upload dialog box open, it's possible to double-click the original widget that spawned it and have another dialog box open
- Clicking on the original window sometimes beeps, sometimes doesn't; focus behaves strangely
- Often after selecting a file, the Open button has no effect... another event needs to be sent to the dialog (clicking inside it, dragging the window, etc.) to get the file to submit

Not sure if these are related, but they all sound like general event handling issues.

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

(In reply to comment #0) 
> - With the file upload dialog box open, it's possible to double-click the
> original widget that spawned it and have another dialog box open
Sounds similar to https://bugzilla.mozilla.org/show_bug.cgi?id=411212#c12
but it was fixed there, I was told.

> - Clicking on the original window sometimes beeps, sometimes doesn't; focus
> behaves strangely
> - Often after selecting a file, the Open button has no effect... another event
> needs to be sent to the dialog (clicking inside it, dragging the window, etc.)
> to get the file to submit
Are these some new regressions? Bug 389931?
Can't reproduce on linux.

Comment 2

[Josh Aas]

Vlad - I can't reproduce anything you listed there on Mac OS X 10.5 using the latest nightly build.

Can you repro reliably? Can you repro with the latest nightly build (I don't know if you're using your own build or what)? What version of Mac OS X are you reproducing on? Are you clicking on the button that opens the file picker or hitting the button with the keyboard?

Smaug - are you able to reproduce anything that Vlad listed in the description?

Comment 3

[Josh Aas]

I can reproduce now - go to a bugzilla add attachment page, open the file picker via a file upload button, then move the dialog out of the way so you can see the button, and click on the button like crazy.

Seems like if you click enough times, perhaps rapidly, events eventually start slipping through. Yuck.

Comment 4

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

1.5.2, latest nightly.

I think Josh and I found an easy way to reproduce:

- Go to the bugzilla form
- click browse
- place mouse back over browse button (with file dialog open) and start clicking like crazy.

Some of the events get through, you'll notice the button highlight.. you can do the same with other widgets, or even just the page itself (e.g. to get text to highlight).  If I do it over the upload control's filename entry field (click like crazy) I can get a new dialog to pop up 100% of the time.

Comment 5

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Oh, if it doesn't trigger quickly (while clicking the browse button, say); move the mouse a pixel or two and keep clicking.

Comment 6

[Steven Michaud [:smichaud] (Retired)]

This bug isn't fixed by the patch for bug 420148 (which was landed
a few days ago -- i.e. it isn't fixed in today's nightly).

But it _is_ fixed by the patch for bug 419668.

Moreover, though the patch for bug 420148 does fix the crash reported
at bug 419668, I think this is just an accident.  The gdb traces that
I'll post in my next two messages show that the patch for bug 420148
doesn't fix the problem that underlies both this bug (bug 420967) and
bug 419668.

Comment 7

[Steven Michaud [:smichaud] (Retired)]

Attachment: Gdb trace of what causes bug 420967 (this bug) and bug 419668 (made on Tiger)

This gdb stack trace shows what causes this bug:  A Cocoa modal event
loop (for an app-modal dialog) is "interrupted" by the processing of a
Gecko event, which in turn leads to a native event being
inappropriately sent (via ProcessNextNativeEvent()) to a ChildView
object outside the app-modal dialog.  This in turn causes the event to
be sent to Gecko (also inappropriately).

This trace is very similar to the one discussed in bug 419668 comment
#25.  Both traces are examples of the same general problem -- bad
things happen when a Cocoa app-modal event loop is "interrupted" by
the processing of a Gecko event, and then native events are allowed to
be processed (via ProcessNextNativeEvent()) "from" that Gecko event.

The patch for bug 419668 appears to fix all cases of this problem.
The patch for bug 420148 clearly does not.

Comment 8

[Steven Michaud [:smichaud] (Retired)]

Attachment: Gdb trace of what causes bug 420967 (this bug) and bug 419668 (made on Leopard)

Comment 9

[Steven Michaud [:smichaud] (Retired)]

Fixed by patch for bug 419668, which just landed.

Comment 10

[Marcia Knous [:marcia]]

verified fixed using Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9pre) Gecko/2008040704 Minefield/3.0pre and the STR in Comment 4. I did some additional testing trying to upload files from various sources including bugzilla and did not encounter any of the issues I had seen previously.

Comment 12

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

This is still happening, though less frequently than before.  Most often manifests itself as the Open button not doing anything until the filepicker window is dragged.

Comment 13

[Steven Michaud [:smichaud] (Retired)]

I doubt very much bug 426363 is the same bug.

But until we have decent STR for bug 426363, I can't know for sure.

And (of course) bug 426363 can't be fixed, either.

Comment 14

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Ah, nice that it was reopened. Meanwhile I'm getting annoyed by this bug in the latest nightly builds. It occurs very often in the "save as" dialog. Pressing Save several times does sometimes nothing. Clicking outside of the dialog executes the action instead.

Could it be that the new issue is a regression which started lately and has nothing to do with the initial comment on this bug?

Comment 15

[Steven Michaud [:smichaud] (Retired)]

Henrik, please open a new bug on your issue and cc me.

Please also try to find reliable STR :-)

Comment 16

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Well, it's entirely possible that this bug fixed the first two issues that I reported; but this other issue is exactly what my third one was about.  So let's reopen 426363 or file a new bug for that.

For STR, just try to pick a file to attach to bugzilla a few dozen times.  You'll hit this eventually; the OK button just won't do anything until you drag the window or click outside the dialog.

Comment 17

[Steven Michaud [:smichaud] (Retired)]

Vlad, I suggest you also open a new bug with all your issues.  It may
turn out to be the same as Henrik's bug ... but we don't know that
yet.

Bug 426363 comment #0 is very vague, and (therefore) not much use.  So
I think it's better to put all your stuff in a new bug.

> For STR, just try to pick a file to attach to bugzilla a few dozen
> times.  You'll hit this eventually; the OK button just won't do
> anything until you drag the window or click outside the dialog.

This isn't a very good STR ... but at least it's fairly specific, and
doesn't sound totally impossible :-)

Cc me on your new bug.

Comment 18

[Henrik Skupin [:whimboo][⌚️UTC+1]]

My seen behavior is exactly the same as what Vlad has written. I can see it on Bugzilla too.

(In reply to comment #17)
> This isn't a very good STR ... but at least it's fairly specific, and
> doesn't sound totally impossible :-)
> 
> Cc me on your new bug.

Me too. Thanks!

Comment 19

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Ok, new bug 473467.