Page 2 of 2

Re: Need help testing PR #1594, which fixes controller issues

Posted: 15 Jan 2018, 02:41
by wareya
The windows nightlies still have not updated.

Re: Need help testing PR #1594, which fixes controller issues

Posted: 15 Jan 2018, 04:14
by Thunderforge
Gotcha, I'm on macOS where the nightly build is still updating regularly. I posted a message on the Windows nightly build thread.

I appreciate your patience in getting this sorted out. It will be good to have another tester once you get your hands on the latest code!

Re: Need help testing PR #1594, which fixes controller issues

Posted: 18 Jan 2018, 04:31
by wareya
I can confirm that the PR didn't break my already-working controller. I have a couple other observations about controller support, though, unrelated to this PR:

1) It's really hard to walk straight forwards/diagonally. Need real 22.5 degree chunk logic, or "analog emulation" like the steam controller does.
2) Analog stick view needs to have a different sensitivity setting than the mouse sensitivity setting, and another sensitivity setting for emulating the cursor on the UI.
3) It seems as though the attack button behavior on triggers is "pressing" at a lower pressure than "releasing". This means that if I press, like, 30%, but the release threshold is at 40%, then even if I release the trigger back to 0% it won't release the attack.

Re: Need help testing PR #1594, which fixes controller issues

Posted: 18 Jan 2018, 21:13
by Thunderforge
Glad to hear it’s still working!

The mappings we’re using are the “third party” SDL_GameControllerDB, so I’m guessing the same problems would also exist with any other games that use that database.

What is your controller? Since it’s likely different than what we’re using, perhaps you could follow their instructions to create a new mapping for your controller and submit it as a PR to the database text file. Then we can update it in OpenMW.

That said, there are likely improvements to controller handling in general that OpenMW can make. Once we rule out that it’s not a controller mapping issue, then I would recommend filing bug reports to get “works, but works badly” issues addressed.

Re: Need help testing PR #1594, which fixes controller issues

Posted: 19 Jan 2018, 07:13
by wareya
1 and 2 don't appear to be mapping-based things.

I have a DualShock 4. I tried looking at the mapping to see what might be causing 3). Its shows:

- lefttrigger:a3
- righttrigger:a4

Which is correct, and mapping should definitely never use the digital button for triggers since it's hypersensitive, but it means that something somewhere else, probably but not necessarily in OpenMW, has incorrect deadzone logic for single-axis analog triggers. In the "wireless controller properties" window (note that my controller is wired, it's just a naming thing), they both start out empty, and fill up moving from left to right as I press them in.

Re: Need help testing PR #1594, which fixes controller issues

Posted: 19 Jan 2018, 18:49
by Thunderforge
Thanks for taking the time to investigate that. Since it doesn’t look like a mapping issue and you seem to have a good understanding of what’s going on, would you mind filing a bug report or two for these issues, since it’s now beyond the scope of the original Mac issue?