Re: My vote for a bug

[Bernard D. <bde...@or...>]

Le 27 mai 08 à 16:54, Bernard Desgraupes a écrit :

>
> Le 24 mai 08 à 16:02, Jon a écrit :
>
>>
>> On May 23, 2008, at 2:00 PM, Joachim wrote:
>>
>>> In OSX 10.5 I cannot type accented letters on an American
>>> keyboard.
>>
>> I observed the same thing just the other day. Likewise with 10.5.2 on
>> an Intel machine.
>>
>
>
> I see this too.
> I just don't understand right now why.
> I looked in the TextInput carbon event: in the
> kEventParamKeyMacCharCodes where normally I would expect the e-acute
> to be, there is only a simple e there after I typed "opt-e + e". If I
> use similar dead keys with a french or with a spanish keyboard, the
> kEventParamKeyMacCharCodes does yield an e-acute (exactly as Joachim
> described). I have no idea why the dead-key and the letter are not
> combined when using the US keyboard. I also tried via the keyboard
> visualizer: when I presse opt-e the keyboard layout is modified so
> that for instance the key for e now displays é, then I press this é
> and still the kEventParamKeyMacCharCodes contains an e.
> I could not find mentions of this problem with Google.

There is indeed a bug in the value provided by the  
kEventParamKeyMacCharCodes parameter in the case of the US keyboard.
I'll ask confirmation on the CarbonDev list but this is very clear  
from a close examination of the data. Possibly a uchr resource is not  
correct or whatever.

Never mind, I have now modified our kEventTextInputUnicodeForKeyEvent  
handler so that it reads directly the Unicode input (found in the  
kEventParamTextInputSendText parameter) and then converts it to  
macRoman.
This fixes the bug and is probably a much better approach.

This will be in AlphaX 8.2a3d4.