Workaround fix for Alt-Gr case....

Workaround fix for Alt-Gr case.

On Windows, "Alt" key will set both alt and meta is on (issue 2215).
But Alt-Gr is used to insert alternate character.

BUG=10846
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=14932

[ukai, 2009-04-30 04:38:38]

I checked I can insert '@' by Ctrl-Alt-2 with Canadian French key layout in
Vista (through RDC/Mac).

I believe this fixes windows regression and doesn't introduce another regression
on linux and mac.

[Hironori Bono, 2009-04-30 06:16:32]

It seems this change still prevents inserting a space character when pressing
control + space keys.
BTW, even though MS drivers don't send ASCII characters when pressing control
keys, it is easy for users to change this behavior on Windows.
So, I'm afraid to say we should not rely on an assumption that keyboards don't
send ASCII characters when pressing a control key on Windows. (This is a reason
why Windows sends three messages for a keyboard event.)
It is better to enclose this block with "#if !defined(OS_WIN)" and "#endif" if
you need to rely on this assumption on Linux and Mac.

[ukai, 2009-04-30 06:31:30]

On 2009/04/30 06:16:32, hbono wrote:
> It seems this change still prevents inserting a space character when pressing
> control + space keys.
> BTW, even though MS drivers don't send ASCII characters when pressing control
> keys, it is easy for users to change this behavior on Windows.
> So, I'm afraid to say we should not rely on an assumption that keyboards don't
> send ASCII characters when pressing a control key on Windows. (This is a
reason
> why Windows sends three messages for a keyboard event.)
> It is better to enclose this block with "#if !defined(OS_WIN)" and "#endif" if
> you need to rely on this assumption on Linux and Mac.

Ok, I understand the situation on Windows.
I enclose this block with "#if !defined(OS_WIN)" for now.
Could you take a look again, please?

Anyway, I'm thinking it might be better to fix this issue in
disambiguateKeyDownEvent or more earlier.  I'll revisit it later.

[Hironori Bono, 2009-04-30 07:02:37]

LGTM.
Thank you for applying my comments.

[ukai, 2009-04-30 07:07:42]

On 2009/04/30 07:02:37, hbono wrote:
> LGTM.
> Thank you for applying my comments.

Thanks for review and valuable information.
Committed revision 14932.
I hope this doesn't cause any regression.
Thanks!