Re: [PATCH 1/3] usbhid: Driver for microsoft natural ergonomic keyboard 4000

[liyu <liyu@ccoss.com.cn>]

    Well, I think you are right.

    I reviewed the "Linux Device Driver", there are some follow words in
that book:

    "the role of a device driver is providing mechanism, not policy."

    It seem my hack break this rule obviously. I will "kcah" it back
soon.   

    Thanks.


Dmitry Torokhov wrote:
> On Sunday 30 July 2006 20:48, liyu wrote:
>   
>> Pavel Machek Wrote:
>>     
>>> Hi!
>>>
>>>   
>>>       
>>>>     This new version get some improvements:
>>>>
>>>>     2. Support left paren key "(", right paren key ")", equal key "=" on
>>>> right-top keypad. In fact, this keyboard generate KEYPAD_XXX usage code
>>>> for them, but I find many applications can not handle them on default
>>>> configuration, especially X.org. To get the most best usability, I use a
>>>> bit magic here: map them to "Shift+9" and "Shift+0".
>>>>     
>>>>         
>>> That is hardly 'improvement'. 'X is broken, so lets break input, too'.
>>>
>>>
>>>
>>>   
>>>       
>> Well, however, this can work truly. If we do not hack as this way.
>> Many applications can not get its input. I think the usability for
>> most people should be first, but not follow rules.
>>
>>     
>
> I do not quite understand why X would have issues with it. KEY_KPEQUAL,
> KEY_KPLEFTPAREN and KEY_KPRIGHTPAREN should work fine even with legacy
> X keyboard driver (one that is using PS/2 protocol instead of evdev).
> You might want to adjust your XKB map or use xmodmap, but kernel should
> report true keycodes.
>  
>   
>> I think we can add one module parameter like "shift_hack" to switch it ?!
>>
>>     
>
> No please don't.
>
>