H
Hisashi Horikawa
Dear,
After upgrading to Windows 10 April 2018 Update (version 1803), behavior of soft keyboard software changed undesirably.
- Up to version 1709, it has behavior as expected
- It is a problem of Windows 10, because it does not depend on the combination of IME and soft keyboard.
In Japan, we use IME to input characters.
When the IME is open, entering keys must add characters to the composition string of the IME.
- Many soft keyboard software simulate key input using the SendInput() API.
- In modern, the KEYEVENTF_UNICODE flag may be specified and the Unicode charactor may be sent.
- In the received application, it is received as VK_PACKET.
Here, after Windows version 1803, the handling of VK_PACKET has been broken for all UWP applications.
As is often the case, if the combination of VK_PACKET and Unicode code point > 0xff come, any charactor is not added to the *composition string*, but is directly entered as a determined string.
This is a very bad behavior in Japan.
1) Conversion to kanji becomes impossible
2) Text can not be entered in the intended order
It seems that there is a problem with dispatching by TSF (Text Services Framework) or processing in DefWindowProc(), or the both.
I strongly hope Microsoft being able to deal promptly.
Regards,
Hisashi.
Continue reading...
After upgrading to Windows 10 April 2018 Update (version 1803), behavior of soft keyboard software changed undesirably.
- Up to version 1709, it has behavior as expected
- It is a problem of Windows 10, because it does not depend on the combination of IME and soft keyboard.
In Japan, we use IME to input characters.
When the IME is open, entering keys must add characters to the composition string of the IME.
- Many soft keyboard software simulate key input using the SendInput() API.
- In modern, the KEYEVENTF_UNICODE flag may be specified and the Unicode charactor may be sent.
- In the received application, it is received as VK_PACKET.
Here, after Windows version 1803, the handling of VK_PACKET has been broken for all UWP applications.
As is often the case, if the combination of VK_PACKET and Unicode code point > 0xff come, any charactor is not added to the *composition string*, but is directly entered as a determined string.
This is a very bad behavior in Japan.
1) Conversion to kanji becomes impossible
2) Text can not be entered in the intended order
It seems that there is a problem with dispatching by TSF (Text Services Framework) or processing in DefWindowProc(), or the both.
I strongly hope Microsoft being able to deal promptly.
Regards,
Hisashi.
Continue reading...