Re: imm32: use SendMessage not PostMessage for the Default IME winproc

[Henri Verbeet]

On 6 November 2012 15:41, Aric Stewart  wrote:
> This allows the calling program to properly get the return codes from the messages when they need them.
> Also prevent all of the calls for IME messages in that windproc from falling back through to DefWindowProc.
>
> for bug 32101

Can we have some tests for this, please? We already have a couple of
known deadlocks involving imm32 message processing, and based on the
amount of tests I'm not sure most of it is based on anything other
than "It seemed like a good idea at the time.".

imm32: use SendMessage not PostMessage for the Default IME winproc

[Aric Stewart]

On 12/11/06 9:38, Henri Verbeet wrote:
> On 6 November 2012 15:41, Aric Stewart  wrote:
> > This allows the calling program to properly get the return codes from the messages when they need them.
> > Also prevent all of the calls for IME messages in that windproc from falling back through to DefWindowProc.
> >
> > for bug 32101
>
> Can we have some tests for this, please? We already have a couple of
> known deadlocks involving imm32 message processing, and based on the
> amount of tests I'm not sure most of it is based on anything other
> than "It seemed like a good idea at the time.".

All right, I have had time to look at this. I have resubmitted with a test showing this change to be correct.

To really get a proper test out of it you will need to run on a windows machine with an IME installed (such as Japanese or Chinese windows) but my tests with my JP XP vm as well as on the test bot shows this to be correct.

Thanks,
-aric