[v4 4/4] winejoystick.drv/joystick_osx.c: adds sim page element controls

[DavidL]

On Tue, Jun 14, 2016 at 12:38 AM, Ken Thomases <ken at codeweavers.com> wrote:

> On Jun 14, 2016, at 1:56 AM, DavidL <david.dljunk at gmail.com> wrote:
>
>
> On Mon, Jun 13, 2016 at 2:46 PM, Ken Thomases <ken at codeweavers.com> wrote:
>
>> Is it intentional that Rudder and Brake both map to the same axis?
>> Later, when enumerating elements, which comes first and thus "wins" the
>> collision (if they ever appear on the same device) is arbitrary based on
>> enumeration order.  That OK?
>>
>
> Yeah I think that's fine as it's unlikely that the same
> joystick/gamepad/controller will have both a listed Rudder and a listed
> Brake axis as one is generally found on a device to control sim-planes and
> the other to control sim-cars. So this is a case where a controller might
> call its RZ axis GD_RZ, Brake, or Rudder depending on the device. Instead
> of GD_X or GD_Y a car controller might call it steering and accelerator.
> Instead of a slider, a joystick might list a throttle. The Linux version
> had some of these too and I thought it better than not just to check in
> case. I realized I did put the wrong link into the comment however. It
> should've been this one:
> https://msdn.microsoft.com/en-us/library/windows/hardware/ff538338(v=vs.85).aspx
>
>
> OK.
>
> By the way, that reminds me: with respect to your commit log, if there's
> anything in there, such as the v2, v3, etc. changes, that shouldn't
> actually be in the ultimate commit log, put it below the "---" line.  The
> Signed-off-by line has to be above that line, though.  So, depending on how
> you're doing things, you may need to use git-format-patch to create files,
> edit those files to put in the notes, and then use git-send-email to send
> those files.  (There's also the git-notes functionality that you can use
> for this, but I haven't spent the time to figure out how to use it.)
>
>
> It seems that treating kIOHIDElementTypeInput_Axis the same as
>> kIOHIDElementTypeInput_Misc is a separate issue from adding support for the
>> Sim page inputs.  Or is there an interrelationship that I'm not aware of?
>> Is it the case that both Axis and Misc elements can be present with either
>> GD or Sim usage pages?
>>
>> Some of the debug logging statements have been changed to say
>> "kIOHIDElementTypeInput_Axis/Misc" while others have been left as
>> "kIOHIDElementTypeInput_Misc".  Is that indicative of something or just an
>> oversight?
>>
>
> Mostly it's because I wasn't sure how the Sim page axes would be listed -
> under Axis or under Misc. For instance, I was very surprised to see the
> Generic_Desktop axes listed under Misc instead of Axis. Do you know why
> that is?
>
>
> No, I don't.  Maybe it's the fact that all of the GD things are treated
> alike and, since some aren't axes, they can't go under Axis.
>
> If the GD_X_Axis is under Misc, not Axis, what's under Axis?
>
>
> I don't know.  I suspect it just depends on how the device manufacturer
> designed the device and its HID reports.
>
> The Mac HID documentation was even more confusing than the Microsoft
> documentation! (at least for me). The changes in the debug statements is
> because the general/default case could now be a Axis or Misc type, but the
> Generic_Desktop page axes are known Misc types.
>
>
> Hmm?  I don't follow how you know that those axes are Misc types (other
> than by the preexisting state of the code, which may not be reliable).
>

You're right, I was just going by the current code and that it seems to
work for my joystick, but I don't actually know that these axes will be
defined as type Misc all the time ... though maybe whoever wrote the
original Mac dinput code (Aric?) knew they should generally/always be
defined as Misc-type somehow? I haven't been able to find any resources on
it. (I've changed all the Traces to Axis/Misc in the latest patch)


>
> So I decided one but not the other. Not changing the Sim page axes debug
> statement on line 496 to Axis/Misc to reflect my lack of knowledge as to
> how they would be listed was indeed an oversight. However, we can go back
> to ignoring the reported Axes elements, but I noticed the dinput code
> didn't ignore them and neither did several other examples of OS X joystick
> programs I found on the web. So I thought, why not just have the axis-type
> fall through, since I'm not actually sure what falls under axis versus
> misc, and try to catch everything? That was my thinking.
>
>
> I think it's good to handle kIOHIDElementTypeInput_Axis.  I just think
> it's a separate issue from the Sim page and so deserves a separate patch.
>
> For what it's worth, the handling in dinput doesn't seem like it could be
> right.  The elements are added to the elements array but then nothing every
> handles them properly after that.  For example, poll_osx_device_state()
> only handles Button and Misc.
>

Yeah I see what you're saying.


>
> -Ken
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.winehq.org/pipermail/wine-devel/attachments/20160614/66230a63/attachment.html>