[ros-dev] [ros-diffs] [dchapyshev] 38663: - Cleanup user32, direct call to kernel space

[Thomas Bluemel]

I also oppose this commit. I like Timo's suggesting of using inline 
functions (not #define). This would ease porting WINE code and would 
provide all the benefits Alex outlined.

Thomas

Aleksey Bragin wrote:
> Please come to a decision. As for my own opinion, there is no need in 
> #defines, because those functions are most probably inlined by the 
> compiler anyway (they are forwarders).
>
>
> On Jan 9, 2009, at 6:26 PM, Alex Ionescu wrote:
>
>> The Windows binaries don't have this behavior, why should ReactOS?
>>
>> The NtUser/Gdi layer should be internal, not called from these sorts 
>> of functions -- it violates principles of private vs public 
>> methods... I know the code isn't OOP but Nt still has strong OOP 
>> design principles at its core.
>>
>> The code in the libraries should, as much as possible, be Win32 code, 
>> not Native API -- ESPECIALLY in the case of NtUser/Gdi.
>>
>> It also makes the binaries MUCH harder to test on Windows, as the 
>> NtUser/Gdi layer in ReactOS Isn't yet compatible with Windows...so 
>> functions would take different parameters, maybe even have different 
>> names.
>>
>> Best regards,
>> Alex Ionescu
>>
>>
>> 2009/1/9 Aleksey Bragin <aleksey at reactos.org 
>> <mailto:aleksey at reactos.org>>
>>
>>     It was not confirmed with me or anyone from Win32k team, however
>>     after talking with Dmitry it looks like there is nothing bad in
>>     it. All those substituted functions are direct forwarders to
>>     win32k both in ReactOS and in Windows.
>>
>>     If James or Timo object, feel free to revert.
>>
>>     WBR,
>>     Aleksey.
>>
>>
>>     On Jan 9, 2009, at 5:49 PM, Ged wrote:
>>
>>>     Lol, I've been through this with Aleksey about 30 minutes ago  ;)
>>>
>>>      
>>>
>>>     Ged.
>>>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Ros-dev mailing list
> Ros-dev at reactos.org
> http://www.reactos.org/mailman/listinfo/ros-dev