<i18n dev> <AWT Dev> Add some missing key maps

Anthony Petrov anthony.petrov at oracle.com
Thu Jun 7 06:12:03 PDT 2012

Hi Charles,

Thanks for pushing this fix. However, we had an internal discussion, and 
here's a few comments regarding it (I'm CC'ing Yuri who provided the 

> On 06/07/2012 10:53 AM, Yuri Nesterenko wrote:
>> (1) XKeysym.java is a generated file; I'd make change in
>> keysym2ucs.h instead, if this routine is not changed, and then
>> run a simple awk script genhash.awk to generate XKeysym.java;
>> (2) Look at the beginning of
>> static Keysym2JavaKeycode getJavaKeycode( long keysym ) --
>> there is a special line for kanaLock added because
>> "Some keysyms maps to more than one keycode".
>> Actually this addition may work OK for Japanese keymap which is
>> usually a single one in XKB setup but MAY totally break e.g. some German
>> or all Croatian. I'd check it first.
> this ^^ getJavaKeycode() method meaning is, kana lock is not missing:
> it is properly returned if we are sure that keyboard _is_ Japanese.
>> (3) Also, regarding shift lock line:
>> look at the code in XToolkit.java dealing with modifiers.
>> Perhaps this line may be risky.

My notes:

1. The generated file issue must certainly be fixed. Charles, could you 
make another patch please?
On a side note, can we just remove the XKeysym.java file altogether and 
simply generate it on each build? Why do we have to keep a generated 
file in the repository?

2 and 3: Charles, please test these issues.

best regards,

On 06/07/12 06:35, Charles Lee wrote:
> Thank you Naoto. The patch has been committed.
> On 06/07/2012 12:45 AM, Naoto Sato wrote:
>> I double checked your patch and confirmed that the table lookup comes
>> after those KANA specific handling, so it looks ok to me.
>> Naoto
>> On 6/5/12 8:13 PM, Charles Lee wrote:
>>> Thank you Naoto. Thank you Anthony.
>>> I do not see the problem. Would some solaris guys take a look on this
>>> issue?
>>> On 06/06/2012 01:56 AM, Naoto Sato wrote:
>>>> The patch is changing the code to always convert VK_KANA_LOCK to
>>>> XK_Kana_Lock keysym. Does this work with Solaris? Looks like there are
>>>> some piece of code that specifically handles KANA_LOCK.
>>>> Naoto
>>>> On 6/5/12 9:02 AM, Anthony Petrov wrote:
>>>>> Hi Charles,
>>>>> I'm not an expert in keyboard-related code, but the code changes look
>>>>> good to me.
>>>>> I think I18n team might want to take a look at the fix as well, so I'm
>>>>> CC'ing the i18n-dev@ mailing list.
>>>>> --
>>>>> best regards,
>>>>> Anthony
>>>>> On 6/5/2012 7:19 AM, Charles Lee wrote:
>>>>>> Hi awt-devs,
>>>>>> There are some key map missing in the Japanese keyboard, for example:
>>>>>> XK_Eisu_Shift, XK_Romaji, etc.
>>>>>> I'd like to propose a patch @
>>>>>> http://cr.openjdk.java.net/~littlee/7174233/webrev.00/
>>>>>> <http://cr.openjdk.java.net/%7Elittlee/7174233/webrev.00/>
>>>>>> The patch is mainly about add these missing maps according to their
>>>>>> means.
>>>>>> Would anyone help to take a look at it?
>>>>>> --
>>>>>> Yours Charles

More information about the i18n-dev mailing list