Re: [PATCH 1/2 v2] user32: Show that mouse move message is generated every time window position is changed

[Dmitry Timoshkov]

Piotr Caban  wrote:
> v2:
> - only patch 2 was changed (fixed detection when message should be
>
> generated)

If only the patch 2 was changed that means that the tests don't really
cover the implementation. That rises the questions: how the logic in the
patch 2 was actually discovered, and what else is missing in the tests
and the implementation?

[PATCH 1/2 v2] user32: Show that mouse move message is generated every time window position is changed

[Piotr Caban]

On 06/08/16 15:49, Dmitry Timoshkov wrote:
> Piotr Caban  wrote:
> > v2:
> > - only patch 2 was changed (fixed detection when message should be
> >
> > generated)
>
> If only the patch 2 was changed that means that the tests don't really
> cover the implementation. That rises the questions: how the logic in the
> patch 2 was actually discovered, and what else is missing in the tests
> and the implementation?

The tests are not covering all combinations of flags passed to 
SetWindowPos (and I think there's no need for that). An example of test 
that will fail with first version of patch is added in patch 2.

Anyway I'll add few more tests for completeness.

[PATCH 1/2 v2] user32: Show that mouse move message is generated every time window position is changed

[Dmitry Timoshkov]

Piotr Caban  wrote:
> > > v2:
> > > - only patch 2 was changed (fixed detection when message should be
> > >
> > > generated)
> >
> > If only the patch 2 was changed that means that the tests don't really
> > cover the implementation. That rises the questions: how the logic in the
> > patch 2 was actually discovered, and what else is missing in the tests
> > and the implementation?
>
> The tests are not covering all combinations of flags passed to 
> SetWindowPos (and I think there's no need for that). An example of test 
> that will fail with first version of patch is added in patch 2.
>
> Anyway I'll add few more tests for completeness.

How about user interactions? For instance when a user drags a window
holding mouse button (without releasing it) over its caption? What
about capture/activation/focus/thread attach states?

[PATCH 1/2 v2] user32: Show that mouse move message is generated every time window position is changed

[Piotr Caban]

On 06/08/16 16:17, Dmitry Timoshkov wrote:
> How about user interactions? For instance when a user drags a window
> holding mouse button (without releasing it) over its caption?

I'm not sure how to test it. The message will be consumed by window 
dragging message loop. Additional mouse message should not affect any 
application.

If mouse is pressed over client area then mouse move message is sent 
correctly with my patch. I'm expecting it to behave the same as clicking 
on window caption on hardware messages level.

What
> about capture/activation/focus/thread attach states?

It also works as expected when mouse is captured. I'll add tests for it. 
There are already tests for activation. I'm not sure what you are asking 
about in case of focus and thread attach states.

Thanks,
Piotr