DOSBox was arbitrarily capping the axes and hats to only what was being emulated.

Wrong, DOSBox intentionally capped mappable features to keep the mapper file
sane in case of replacing joysticks or switching to keyboard-only mapping of joystick
targets and back to real joysticks.

I don't have any joystick any more so hope somebody is interested in checking out
ways of testing this or a more useful implementation rather than "arbitrarily removing
the arbitrary capping".

Hey wd, I don't understand the slightly hostile tone in this matter. There are a few threads about joystick troubles connected to the "axis limit" IIRC, and I remember not being able to use the right thumbstick on my Xbox 360 controller (XBCD driver) at all. I don't have the knowledge (and motivation/time, to be frank) to setup a compile environment. If someone could compile a Windows build with that patch applied I could do some quick Windows 7 testing and reporting though if that's any help.

Wrong, DOSBox intentionally capped mappable features to keep the mapper file sane in case of replacing joysticks or switching to keyboard-only mapping of joystick targets and back to real joysticks.

I'm trying to think of a situation where the mapper file would stop being sane if DOSBox recognized 8 axes instead of just limiting itself to 4 axes... and why it would stop being sane if DOSBox recognized a hat for binding even when not emulating a hat...

I'm failing to see how the addition of about 16 more possible bindable inputs (4 axes * 2 directions + 2 hats * 4 directions) would have any impact on the sanity of the file, when the keyboard itself has about 101...

DOSBox without this patch is literally throwing away inputs. Sure if you never have more than 4 axes on your joystick, you would never notice, but Microsoft pumped out a controller with 5 axes, and arranged them somewhat retardedly: Axes 0,1 = Left control stick, Axis 2 = the triggers together as one axis, Axes 3,4 = Right control stick. Put this together with DOSBox arbitrarily capping at 4 axes, and you only get one-axis of the right control stick, with no possible way to map Axes 3,4 onto emulated Axes 2,3.

I don't have any joystick any more so hope somebody is interested in checking out ways of testing this or a more useful implementation rather than "arbitrarily removing the arbitrary capping".

There is no way to implement this except removing the arbitrary cap on input (or rather raising the arbitrary cap on input... if someone had a controller with more than 8-axes, it would now still throw away the 9th axis and above).

As for testing, I'm all for testing anything you can possibly dream up that you think I might be forgetting about... but as is, it's all the same code that has been working for years... just reading more axes and the hat even if there isn't a hat being emulated.

@ADDiCT attached, I have the Windows binary I built from the raw SVN tarball with patch applied.

Wide testing base == good.

I don't understand the slightly hostile tone in this matter

Why would I care whether you understand or not?

Because you are a caring and friendly person?

@cfoesch A BIG THANK YOU, i was very dissapointed that my controller did not fully worked (Hat & right thumb does not worked before), Now i can play like it should be, Thank you very much again, i hope this is included into the next release, if not then please continue to compile your awesome build. THANKS.

Taewoong's SVN build has added this patch. I'll try my logitech F710 in X360 mode to test it's effect.

http://ykhwong.x-y.net/

mgtroyas wrote:

Taewoong's SVN build has added this patch. I'll try my logitech F710 in X360 mode to test it's effect.

http://ykhwong.x-y.net/

Yeah, i asked him to add this patch, i'm very glad he did it, i don't have experienced any trouble during this 2 months of use. 😊

If there's no cap on buttons then why cap axes and hats? I could use a dualshock 3 with both left and right stick because the rest of the controller was mapped to buttons, no analog triggers or hat. I could map up to 17 buttons with no issues (with button wrap off), even though dosbox was only emulating a max number of 6 buttons with the ch profile. For some reason this same approach isn't applied to axes. If dosbox is only emulating 4 axes you can only use 4, and the bigger problem is that you can't even tell it which 4 to use. It will just choose the 4 itself which is what messes up the 360 controller. At the very least you should be able to drop the analog triggers in favor of another axis on the right stick if you still want to cap 4 axes.

I'm pretty new at using DosBox and haven't even begun to even consider attempting to foray into its source to try and make changes, so I just wanted to extend my gratitude for providing this change and the accompanying build. I can now play MechWarrior 2 using both Xbox 360 joysticks (whereas before I couldn't bind anything to the X axis of the right joystick). I'm kind of ecstatic now and am gonna spend the rest of the day in pure nostalgic bliss.

@cfoesch, thank you!!!

Will this excellent enhancement work in Linux dosbox?

Thanks...

I would like to know, how I can interact with this patch (config options). I am using SVN-Daum, but I find no instructions, on what to do. I have a Logitech Rumblepad F710, which I got to get working with X-Wing and TIE-Fighter, but gives me headaches with Strike Commander. I can use the Joypad in the menu, but I get no functionality ingame. Whatever I chose, no luck. I also tried using it in DirectInput mode and XInput mode. Does it count, which input mode I use? I guess, since this patch was made for the XBox controller, originally, it should be XInput (the Joypad I am using can be switched to either)?

So, after having this patch applied (i.e SVN-Daum installed), what is the next step do do? Thanks a lot!

Andreas

cfoesch wrote:

@ADDiCT attached, I have the Windows binary I built from the raw SVN tarball with patch applied.

Wide testing base == good.

Thank you cfoesch. I've been using this SVN build of yours for ages now. Works great with DS3 and move controller on ScpToolkit.

Would it be possible to ask of you a similar Windows binary based on DOSBox 0.74-2 with your patch applied?

InterClaw wrote:

Would it be possible to ask of you a similar Windows binary based on DOSBox 0.74-2 with your patch applied?

You realized that the thread was dead for almost 5 years now, did you? 😉 Oh, and cfoesch hasn't been active here since 2011, btw.

However, my ECE build (see my signature) has his patch integrated and is based on the latest SVN, go give that one a try!

Thanks Yesterplay80! Your version is great. 😀

Yeah, it was kind of a longshot since the thread was so old. But now I have your version instead!

Anyone have this patch that works with more recent SVN source please?