xorg/driver/xf86-input-fpit - [PATCH] Don't call xf86IsCorePointer.

[pcpa at mandriva.com.br]

Quoting Peter Hutterer <mailinglists at who-t.net>:

>> http://www.nabble.com/Bug-457722%3A-xserver-xorg-input-fpit%3A-using-touchscreen-kills-x-td14492063.html
>
> yeah, this is bug #14057 on bugs.fdo.org.
> i just noticed that I did attach some patches to this bug that may 
> fix things up.

  I did some more searches, and did not find anything useful in the
first 10 google pages. So I believe there isn't much people using
X Server 1.4 with the new input api.

>> Since it was calling xf86IsCorePointer in a loop, I did a different
>> patch (maybe I just did the compiler optimization by removing the set

  Actually, I think I misunderstood what you asked, it should not
be required to call xf86IsCorePointer inside the loop (it shouldn't
change during that small loop), but that afects only the old api, and
xf86IsCorePointer is a cheap call, basically check if the argument
pointer is the same as inputInfo.pointer.

>> the variable to 0 every iteration...). The loop is a bit weird also,
>> as it is basically:
>> for (loop = 0; loop < end; loop++) {
>>    ...
>>    loop += chunk_size - 1;
>> }
>> but I did not modify it to avoid the need of a bigger patch.
>
> If you have the desire to fix it, feel free to do so. I'll do my best 
> reviewing but unless we hear back from an actual user of such a 
> device, there isn't all that much we can do.

  I think it is not required to modify it. Would be basically only an
aesthetic change. And I already did one by adding braces to the nested
if, i.e. "if (c1) if (c2) s;" to "if (c1) { if (c2) s; }", and the second
if is also "#if'ed" out anyway for the current input api.

> Cheers,
>   Peter

Paulo