Qmk_firmware: Georgi NKRO fallback mode sends backspace for the ‘I’ brief

[fedde-s]

I'm running into an issue with my Georgi stenographic keyboard when using it in ‘fake steno/NKRO fallback’ mode. A common steno stroke for the pronoun ‘I’, the chord RE | RU, sends a Backspace instead of simulating the chording of N and M on a QWERTY keyboard.

I haven't read the code in too much detail, but I think it might have something to do with the else after this if condition. It seems to refer back to the same logic as the QWERTY/gaming mode, in which that chord is indeed supposed to send a Backspace.

System Information

• Keyboard: Georgi
  
  
  
  • Revision (if applicable): Rev2
  
  
• Operating system: reproduced on both Debian GNU/Linux and Android
• QMK Firmware version: v1.1: ClayMAug 4 2019 (according to the #-TSDZ chord)
• Any keyboard related software installed?
  
  
  
  • [ ] AutoHotKey
  
  
  • [ ] Karabiner
  
  
  • [x] Other: Plover for Linux/Dotterel for Android
  
  

[fedde-s]

I'm sorry if this is not the right place to summarise the issue. But if it is: @germ, do you see what the problem I'm running into is? And if so, given your familiarity with the code, does a solution spring to mind that doesn't do too much harm to its maintainability?

[germ]

The problem is that the parsing code is calling the main dictionary instead of the proper NKRO fallback when trying to parse out chords. In the mean time, try and use the serial interface with Plover.

I'll have a PR ready for the qmk team in a few hours and link it :)

[fedde-s]

In the mean time, try and use the serial interface with Plover.

That mode definitely works great for me so far! And even when it's unavailable, such as on Android, a single-letter word isn't too hard to fingerspell if I remember to use two extra fingers 😅

[fedde-s]

I later noticed that the NKRO/fake steno fallback mode also sends only the bottommost key, the QWERTY home row key, when I hit a chord consisting solely of two consonant keys above each other (such as PW for “about”). If I understand your diagnosis of the problem correctly, this is caused by the same bug.

[fedde-s]

@germ, is this still in progress? Is there anything I could help with when I find the time to read up on how to build and load changed code to test it in practice, or do you think that would only make it harder? Not urgently waiting for it, but I'm curious to learn more and contribute.

[germ]

Oh crap, forgot about this. I don't have my Georgi right now but I'll try hacking at it in the next day or two. I'll ping the Plover discord good chance one of them can fix it up in 5 seconds :)

(Wrote it on my hand so hopefully I don't forget this time)

[drashna]

May need to update the matrix too :D