Hi, Oleg,
I don't see anything wrong with the fix (which hopefully means that it's
fine :))
One cosmetic issue (no need to send a new version for review): could you
change ImmGetContext() to ::ImmGetContext(), so it's easy to distinguish
between AwtComponent's methods and Win32 calls, please?
Thanks,
Artem
On 6/22/2012 8:28 PM, Oleg Pekhovskiy wrote:
Hi,
Please review the second version of fix for CR:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7024749
Webrev:
http://cr.openjdk.java.net/~bagiras/8/7024749.2
It resolves applet's IME typing problems that existed with the first fix.
Here are my answers to Artem's comments:
1. I see no reason to change ImmGetContext() to ImmGetHWnd(). Everywhere
in the code ImmGetHWnd() is followed by ImmGetContext(), and these two
calls can easily be combined into a single method in AwtComponent.
ImmGetHWnd() & ImmGetContext() return HWND and HIMC accordingly that are
used BOTH in ImmReleaseContext().
2. Comment about focus proxy in AwtComponent::OpenCandidateWindow() is
now obsolete. Instead, you need to add a comment why we send
WM_IME_NOTIFY to this component (GetHWnd()), not to its focus proxy.
It's not obsolete now, because I returned GetProxyFocusOwner() there.
3. There is no need to call ImmReleaseContext() if hIMC is NULL in
AwtComponent::WmImeSetContext().
Thanks, I changed that.
4. In WM_ACTIVATE handler, I would first handle WM_IME_ENDCOMPOSITION,
then call ImmReleaseContext().
Seems like it would be better to release IMM Context for avoiding its
simultaneous usage.
5. In the same WM_ACTIVATE handler, what's the reason of direct call to
DefWindowProc() instead of regular ::SendMessage()?
It's not my code, so I left that as is.
PS: this fix leads to the regression for
'test/java/awt/Focus/AppletInitialFocusTest', but I was not able to fix
that yet (ready to file a separate CR).
Thanks,
Oleg
