Touch-initiated drag-out to download file but fails.

Touch-initiated drag-out to download file but fails.

BUG=

[Hongbo Min, 2013-04-16 09:26:42]

I add some logging output to track the issue in drag-out downloading file.

The following is the output on my Win8 tablet after long press on the file link
at http://www.thecssninja.com/demo/gmail_dragout/:

- call WebDragSource::ShouldDropDragSource: key_state: 2"
- call WebContentsDragWin::OnWaitForData"

Then the browser window gets freezed, no responsiveness.

The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask, and is
expected to be continued since the DragSourceWin::QueryContinueDrag returns
S_OK.

Any hints?

[dcheng, 2013-04-16 21:01:17]

On 2013/04/16 09:26:42, Hongbo Min wrote:
> I add some logging output to track the issue in drag-out downloading file.
> 
> The following is the output on my Win8 tablet after long press on the file
link
> at http://www.thecssninja.com/demo/gmail_dragout/:
> 
> - call WebDragSource::ShouldDropDragSource: key_state: 2"
> - call WebContentsDragWin::OnWaitForData"
> 
> Then the browser window gets freezed, no responsiveness.
> 
> The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask, and is
> expected to be continued since the DragSourceWin::QueryContinueDrag returns
> S_OK.
> 
> Any hints?

I'm going to try to find a touch device to test on locally. If I had to guess,
there's a deadlock somewhere =(

[Hongbo Min, 2013-04-19 13:25:04]

On 2013/04/16 21:01:17, dcheng wrote:
> On 2013/04/16 09:26:42, Hongbo Min wrote:
> > I add some logging output to track the issue in drag-out downloading file.
> > 
> > The following is the output on my Win8 tablet after long press on the file
> link
> > at http://www.thecssninja.com/demo/gmail_dragout/:
> > 
> > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > - call WebContentsDragWin::OnWaitForData"
> > 
> > Then the browser window gets freezed, no responsiveness.
> > 
> > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask, and
is
> > expected to be continued since the DragSourceWin::QueryContinueDrag returns
> > S_OK.
> > 
> > Any hints?
> 
> I'm going to try to find a touch device to test on locally. If I had to guess,
> there's a deadlock somewhere =(

dcheng@, how is going on? Do you have any progress?

[dcheng, 2013-04-19 20:24:12]

On 2013/04/19 13:25:04, Hongbo Min wrote:
> On 2013/04/16 21:01:17, dcheng wrote:
> > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > I add some logging output to track the issue in drag-out downloading file.
> > > 
> > > The following is the output on my Win8 tablet after long press on the file
> > link
> > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > 
> > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > - call WebContentsDragWin::OnWaitForData"
> > > 
> > > Then the browser window gets freezed, no responsiveness.
> > > 
> > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask, and
> is
> > > expected to be continued since the DragSourceWin::QueryContinueDrag
returns
> > > S_OK.
> > > 
> > > Any hints?
> > 
> > I'm going to try to find a touch device to test on locally. If I had to
guess,
> > there's a deadlock somewhere =(
> 
> dcheng@, how is going on? Do you have any progress?

Sorry about the delay. I had some issues setting up Win8 but I plan on testing
this today.

[dcheng, 2013-04-23 04:04:20]

On 2013/04/19 20:24:12, dcheng wrote:
> On 2013/04/19 13:25:04, Hongbo Min wrote:
> > On 2013/04/16 21:01:17, dcheng wrote:
> > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > I add some logging output to track the issue in drag-out downloading
file.
> > > > 
> > > > The following is the output on my Win8 tablet after long press on the
file
> > > link
> > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > 
> > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > - call WebContentsDragWin::OnWaitForData"
> > > > 
> > > > Then the browser window gets freezed, no responsiveness.
> > > > 
> > > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask,
and
> > is
> > > > expected to be continued since the DragSourceWin::QueryContinueDrag
> returns
> > > > S_OK.
> > > > 
> > > > Any hints?
> > > 
> > > I'm going to try to find a touch device to test on locally. If I had to
> guess,
> > > there's a deadlock somewhere =(
> > 
> > dcheng@, how is going on? Do you have any progress?
> 
> Sorry about the delay. I had some issues setting up Win8 but I plan on testing
> this today.

Sorry I finally managed to test this more thoroughly today. I'm pretty sure this
is the bug I linked earlier (http://crbug.com/163387). We should probably fix
that...

[dcheng, 2013-04-23 04:17:32]

On 2013/04/23 04:04:20, dcheng wrote:
> On 2013/04/19 20:24:12, dcheng wrote:
> > On 2013/04/19 13:25:04, Hongbo Min wrote:
> > > On 2013/04/16 21:01:17, dcheng wrote:
> > > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > > I add some logging output to track the issue in drag-out downloading
> file.
> > > > > 
> > > > > The following is the output on my Win8 tablet after long press on the
> file
> > > > link
> > > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > > 
> > > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > > - call WebContentsDragWin::OnWaitForData"
> > > > > 
> > > > > Then the browser window gets freezed, no responsiveness.
> > > > > 
> > > > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN mask,
> and
> > > is
> > > > > expected to be continued since the DragSourceWin::QueryContinueDrag
> > returns
> > > > > S_OK.
> > > > > 
> > > > > Any hints?
> > > > 
> > > > I'm going to try to find a touch device to test on locally. If I had to
> > guess,
> > > > there's a deadlock somewhere =(
> > > 
> > > dcheng@, how is going on? Do you have any progress?
> > 
> > Sorry about the delay. I had some issues setting up Win8 but I plan on
testing
> > this today.
> 
> Sorry I finally managed to test this more thoroughly today. I'm pretty sure
this
> is the bug I linked earlier (http://crbug.com/163387). We should probably fix
> that...

(By we, I guess I mean "I").

[dcheng, 2013-04-24 05:38:56]

On 2013/04/23 04:17:32, dcheng wrote:
> On 2013/04/23 04:04:20, dcheng wrote:
> > On 2013/04/19 20:24:12, dcheng wrote:
> > > On 2013/04/19 13:25:04, Hongbo Min wrote:
> > > > On 2013/04/16 21:01:17, dcheng wrote:
> > > > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > > > I add some logging output to track the issue in drag-out downloading
> > file.
> > > > > > 
> > > > > > The following is the output on my Win8 tablet after long press on
the
> > file
> > > > > link
> > > > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > > > 
> > > > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > > > - call WebContentsDragWin::OnWaitForData"
> > > > > > 
> > > > > > Then the browser window gets freezed, no responsiveness.
> > > > > > 
> > > > > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN
mask,
> > and
> > > > is
> > > > > > expected to be continued since the DragSourceWin::QueryContinueDrag
> > > returns
> > > > > > S_OK.
> > > > > > 
> > > > > > Any hints?
> > > > > 
> > > > > I'm going to try to find a touch device to test on locally. If I had
to
> > > guess,
> > > > > there's a deadlock somewhere =(
> > > > 
> > > > dcheng@, how is going on? Do you have any progress?
> > > 
> > > Sorry about the delay. I had some issues setting up Win8 but I plan on
> testing
> > > this today.
> > 
> > Sorry I finally managed to test this more thoroughly today. I'm pretty sure
> this
> > is the bug I linked earlier (http://crbug.com/163387). We should probably
fix
> > that...
> 
> (By we, I guess I mean "I").

After some more analysis, the issue is at
http://src.chromium.org/viewvc/chrome/trunk/src/ui/base/dragdrop/os_exchange_....

As a hack to test your solution, change the check to also check for VK_RBUTTON.
I think I have a better permanent solution though.

[Hongbo Min, 2013-04-24 05:43:21]

On 2013/04/24 05:38:56, dcheng wrote:
> On 2013/04/23 04:17:32, dcheng wrote:
> > On 2013/04/23 04:04:20, dcheng wrote:
> > > On 2013/04/19 20:24:12, dcheng wrote:
> > > > On 2013/04/19 13:25:04, Hongbo Min wrote:
> > > > > On 2013/04/16 21:01:17, dcheng wrote:
> > > > > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > > > > I add some logging output to track the issue in drag-out
downloading
> > > file.
> > > > > > > 
> > > > > > > The following is the output on my Win8 tablet after long press on
> the
> > > file
> > > > > > link
> > > > > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > > > > 
> > > > > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > > > > - call WebContentsDragWin::OnWaitForData"
> > > > > > > 
> > > > > > > Then the browser window gets freezed, no responsiveness.
> > > > > > > 
> > > > > > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN
> mask,
> > > and
> > > > > is
> > > > > > > expected to be continued since the
DragSourceWin::QueryContinueDrag
> > > > returns
> > > > > > > S_OK.
> > > > > > > 
> > > > > > > Any hints?
> > > > > > 
> > > > > > I'm going to try to find a touch device to test on locally. If I had
> to
> > > > guess,
> > > > > > there's a deadlock somewhere =(
> > > > > 
> > > > > dcheng@, how is going on? Do you have any progress?
> > > > 
> > > > Sorry about the delay. I had some issues setting up Win8 but I plan on
> > testing
> > > > this today.
> > > 
> > > Sorry I finally managed to test this more thoroughly today. I'm pretty
sure
> > this
> > > is the bug I linked earlier (http://crbug.com/163387). We should probably
> fix
> > > that...
> > 
> > (By we, I guess I mean "I").
> 
> After some more analysis, the issue is at
>
http://src.chromium.org/viewvc/chrome/trunk/src/ui/base/dragdrop/os_exchange_....
> 
> As a hack to test your solution, change the check to also check for
VK_RBUTTON.
> I think I have a better permanent solution though.

Thanks for your sharing. Could you please also share your permanent solution or
a patch set? I am interesting in how to make DoDragDrop happy on touch screen.

[dcheng, 2013-04-24 05:55:48]

On 2013/04/24 05:43:21, Hongbo Min wrote:
> On 2013/04/24 05:38:56, dcheng wrote:
> > On 2013/04/23 04:17:32, dcheng wrote:
> > > On 2013/04/23 04:04:20, dcheng wrote:
> > > > On 2013/04/19 20:24:12, dcheng wrote:
> > > > > On 2013/04/19 13:25:04, Hongbo Min wrote:
> > > > > > On 2013/04/16 21:01:17, dcheng wrote:
> > > > > > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > > > > > I add some logging output to track the issue in drag-out
> downloading
> > > > file.
> > > > > > > > 
> > > > > > > > The following is the output on my Win8 tablet after long press
on
> > the
> > > > file
> > > > > > > link
> > > > > > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > > > > > 
> > > > > > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > > > > > - call WebContentsDragWin::OnWaitForData"
> > > > > > > > 
> > > > > > > > Then the browser window gets freezed, no responsiveness.
> > > > > > > > 
> > > > > > > > The fact that key_state is 2 means DoDragDrop got MK_RBUTTONDOWN
> > mask,
> > > > and
> > > > > > is
> > > > > > > > expected to be continued since the
> DragSourceWin::QueryContinueDrag
> > > > > returns
> > > > > > > > S_OK.
> > > > > > > > 
> > > > > > > > Any hints?
> > > > > > > 
> > > > > > > I'm going to try to find a touch device to test on locally. If I
had
> > to
> > > > > guess,
> > > > > > > there's a deadlock somewhere =(
> > > > > > 
> > > > > > dcheng@, how is going on? Do you have any progress?
> > > > > 
> > > > > Sorry about the delay. I had some issues setting up Win8 but I plan on
> > > testing
> > > > > this today.
> > > > 
> > > > Sorry I finally managed to test this more thoroughly today. I'm pretty
> sure
> > > this
> > > > is the bug I linked earlier (http://crbug.com/163387). We should
probably
> > fix
> > > > that...
> > > 
> > > (By we, I guess I mean "I").
> > 
> > After some more analysis, the issue is at
> >
>
http://src.chromium.org/viewvc/chrome/trunk/src/ui/base/dragdrop/os_exchange_....
> > 
> > As a hack to test your solution, change the check to also check for
> VK_RBUTTON.
> > I think I have a better permanent solution though.
> 
> Thanks for your sharing. Could you please also share your permanent solution
or
> a patch set? I am interesting in how to make DoDragDrop happy on touch screen.

I need to try some experiments when I'm in the office tomorrow for the permanent
fix for the deadlock, but I will try to have it checked in by 5PM PST tomorrow.

As for your original patch, I think it's largely OK. I have a few minor
comments, but I was wondering if you wanted to update it to include the support
for dragging out DownloadURL before I look at it again?

Sorry for the long investigation; I had some issues getting machines and setting
them up. =/

[Hongbo Min, 2013-04-24 06:03:25]

On 2013/04/24 05:55:48, dcheng wrote:
> On 2013/04/24 05:43:21, Hongbo Min wrote:
> > On 2013/04/24 05:38:56, dcheng wrote:
> > > On 2013/04/23 04:17:32, dcheng wrote:
> > > > On 2013/04/23 04:04:20, dcheng wrote:
> > > > > On 2013/04/19 20:24:12, dcheng wrote:
> > > > > > On 2013/04/19 13:25:04, Hongbo Min wrote:
> > > > > > > On 2013/04/16 21:01:17, dcheng wrote:
> > > > > > > > On 2013/04/16 09:26:42, Hongbo Min wrote:
> > > > > > > > > I add some logging output to track the issue in drag-out
> > downloading
> > > > > file.
> > > > > > > > > 
> > > > > > > > > The following is the output on my Win8 tablet after long press
> on
> > > the
> > > > > file
> > > > > > > > link
> > > > > > > > > at http://www.thecssninja.com/demo/gmail_dragout/:
> > > > > > > > > 
> > > > > > > > > - call WebDragSource::ShouldDropDragSource: key_state: 2"
> > > > > > > > > - call WebContentsDragWin::OnWaitForData"
> > > > > > > > > 
> > > > > > > > > Then the browser window gets freezed, no responsiveness.
> > > > > > > > > 
> > > > > > > > > The fact that key_state is 2 means DoDragDrop got
MK_RBUTTONDOWN
> > > mask,
> > > > > and
> > > > > > > is
> > > > > > > > > expected to be continued since the
> > DragSourceWin::QueryContinueDrag
> > > > > > returns
> > > > > > > > > S_OK.
> > > > > > > > > 
> > > > > > > > > Any hints?
> > > > > > > > 
> > > > > > > > I'm going to try to find a touch device to test on locally. If I
> had
> > > to
> > > > > > guess,
> > > > > > > > there's a deadlock somewhere =(
> > > > > > > 
> > > > > > > dcheng@, how is going on? Do you have any progress?
> > > > > > 
> > > > > > Sorry about the delay. I had some issues setting up Win8 but I plan
on
> > > > testing
> > > > > > this today.
> > > > > 
> > > > > Sorry I finally managed to test this more thoroughly today. I'm pretty
> > sure
> > > > this
> > > > > is the bug I linked earlier (http://crbug.com/163387). We should
> probably
> > > fix
> > > > > that...
> > > > 
> > > > (By we, I guess I mean "I").
> > > 
> > > After some more analysis, the issue is at
> > >
> >
>
http://src.chromium.org/viewvc/chrome/trunk/src/ui/base/dragdrop/os_exchange_....
> > > 
> > > As a hack to test your solution, change the check to also check for
> > VK_RBUTTON.
> > > I think I have a better permanent solution though.
> > 
> > Thanks for your sharing. Could you please also share your permanent solution
> or
> > a patch set? I am interesting in how to make DoDragDrop happy on touch
screen.
> 
> I need to try some experiments when I'm in the office tomorrow for the
permanent
> fix for the deadlock, but I will try to have it checked in by 5PM PST
tomorrow.
> 
> As for your original patch, I think it's largely OK. I have a few minor
> comments, but I was wondering if you wanted to update it to include the
support
> for dragging out DownloadURL before I look at it again?
> 
> Sorry for the long investigation; I had some issues getting machines and
setting
> them up. =/

OK, thanks for your further deep investigation. I'd like to update it and merge
the drag-out downloading to the CL https://codereview.chromium.org/14122008/.