georgi Tag

9 Articles


26 Aug 2021

categories: keebs steno
tags: ergodox georgi daisy mb44 fruitbar

I’d like to take a moment to rebase where I’m at with steno (and keyboards), and provide a little context. At the time of publishing this, the blog website has just gone live, and all prior articles from my steno journal have been imported and reformatted.

First things first: I have not done much practice with steno in 2021. I have done quite a bit of practice building mechanical (qwerty) keyboards though. I’ve also headed off into the weeds with some raspberry-Pi-based projects that I can’t talk about yet.

A brief inventory of my keyboards:

  • I use an Ergodox EZ every day at work. I keep it in qwerty mode, but in the past I’ve tinkered with putting steno in its firmware.
  • I have a Georgi that’s all souped up with symbol and media layers, as well as a dictionary of nav commands.
  • I’ve experimented with split boards smaller than the Ergodox (Kailh Choco, pinky3, and Sofle). I really wanted something in a smaller profile, and with rotary encoders. I had a lot of fun building them but in the end the thumb positions actually started causing my first RSI! I also found that I have grown used to the huge number of keys available on the Ergodox.
  • I’ve made some forays into the land of 40%, with Daisy, JD40 MkII, and TG4X. This experiment has stalled, as I’m currently waiting on a group buy for the MB-44.
  • I’ve thought long and hard about a 4-row 50% keyboard, and then in my research found that others are currently making them so I don’t have to. As of this writing I eagerly await the group buy for the Fruitbar. I expect that once I have a more office-centric job I’ll take the Ergodox there, and at home I’ll use this longboi keyboard.

As far as my steno work goes…it’s been a while since I’ve touched it. I still would love to get back to it, moreso now even due to my new RSI.
Perhaps I’ll write a proper breakdown of how that machine works for me, and why I built the firmware and dictionaries the way I did.


24 Nov 2020

categories: steno
tags: ergodox georgi

Made a cheat-sheet for the single-stroke commands dictionary tonight. Did my first cut, paste and undo! The shapes are a little goofy but I think they’ll suffice for my needs. Might even allow me to free up some keys on the firmware layer (no need for arrows, letters, or functions).
I did order an Ergodox. The clutter argument was very compelling. It’ll take 4 weeks to get here, so in the meantime I’ll keep practicing on Georgi.
I’m excited to put together a firmware for Ergodox that does everything I need–and with LED control!


23 Nov 2020

categories: keebs steno
tags: qmk butterstick georgi ergodox

A few things I noticed in practice tonight

  • tapping the OSL key doesn’t let me alt-tab with impunity, but holding it does.

  • my custom nav layer doesn’t support select-all, cut, copy, paste, undo, or redo. I wonder again if I should record the number of times I use a feature in a day, and put those at the front and center.
    oh, also ctrl-backspace (or alt-bsp and cmd-bsp on mac).
    Even if I put my most common commands on the first nav layer, I still need either

    1. (hw, current status) a qwerty keyboard standing by.
      pro: I have it (it’s free)
      con: slow to switch from typing to commanding
      con: BIG desk clutter
      con: need the qwerty keyboard for typing while learning steno
    2. (hw) a macropad (e.g. Butterstick) standing by.
      pro: I have it (it’s free)
      con: slow to switch from typing to commanding
      con: smol desk clutter
      con: need the qwerty keyboard for typing while learning steno
    3. (fw) three or more layers on the georgi.
      pro: I have it (it’s free)
      pro: everything is at my fingertips (it’s quick to switch)
      pro: no clutter!
      con: patience and time to tweak and learn the layers
      con: need the qwerty keyboard for typing while learning steno
    4. (hw) A BIGGER georgi, with more keys to hand (see: Gergo. I think maybe Gergoplex isn’t enough columns).
      pro: it’s a kit I can build myself (another round of custom switches, this time maybe kailh silver?)
      pro: everything is at my fingertips (quick to switch)
      pro: no clutter!
      pro: less need to layer the commands (quicker than switching)
      pro: do not need a qwerty kicking around while I learn steno
      con: hard to program it to be steno sometimes and command sometimes?
      con: I’d need to build my own enclosure for it.
    5. (hw) white tiger option: get an ergodox already (~$350).
      pro: it’s a kit I can configure myself (it does hotswap)
      pro: everything is at my fingertips
      pro: no clutter!
      pro: less need to layer the commands
      pro: do not need a qwerty kicking around while I learn steno
      con: hard to program it to be steno sometimes and command sometimes?
      con: expensive
    6. (sw) learn and develop the Single Stroke commands dictionary in Plover.
      pro: I have it (it’s free)
      pro: VERY easy to reconfigure a broken brief on the fly
      con: need the qwerty keyboard for typing while learning steno
      con: patience and time to tweak and learn the briefs

Options 4 and 5 (Gergo and Ergodox) are similar enough that I can runoff them: With Gergo, I have to design and build the enclosure, and buy keycaps and switches. With Ergodox, I just have to spend ~$280 marginally (and maybe make or buy steno-friendly keycaps). With Gergo, I get to make it look however I want, with Ergodox, no such personalization.

+ personalized  - cost
- kit work      - not personalized

Actually, now that I think about it, the Gergo wouldn’t be a full size qwerty, so I’d be training (1) steno, (2) columnar qwerty, and (3) gergo commands and shortcuts for qwerty. Ergodox loses on customization, but “costs” less in terms of money-work, and wins on me not doing that (3) training.

I’d also point out that the “do not need the qwerty keyboard for typing while learning steno” is not worth nothing.

In the end it’s really down to an ideological choice: do I want to do this retraining of my brain and fingers through software, firmware, or hardware? And even if I do it through software, do I want the greater hardware anyway to speed things along, declutter the process, and/or be more ergonomic for me? (For the record I have yet to experience any RSI…)

I will sleep on it.


22 Nov 2020

categories: steno
tags: qmk butterstick georgi

Last night / this morning I read a lot about how QMK works with layers. Then I worked up keymaps for both the Butterstick and the Georgi!
The first milestone is complete: I can successfully alt-tab any number of times!

Next milestones:

  • make the nav layer key (and fn layer key) momentary-layers (like OSL, but that didn’t seem to work immediately).

    OSL(layer) - momentarily activates layer until the next key is pressed. See One Shot Keys for details and additional functionality.

    // One-shot layer - 256 layer max
    #define OSL(layer) (QK_ONE_SHOT_LAYER | ((layer)&0xFF))

    Defined in quantum/quantum_keycodes.h and used in the action_for_keycode function of quantum/keymap_common.c.

            // OSL(action_layer) - One-shot action_layer
            action_layer = keycode & 0xFF;
            action.code  = ACTION_LAYER_ONESHOT(action_layer);
        case QK_ONE_SHOT_MOD ... QK_ONE_SHOT_MOD_MAX:;
            // OSM(mod) - One-shot mod
            mod         = mod_config(keycode & 0xFF);
            action.code = ACTION_MODS_ONESHOT(mod);

    which leads to

    #define ACTION_LAYER_TAP(layer, key) ACTION(ACT_LAYER_TAP, (layer) << 8 | (key))
    num action_kind_id { ACT_LAYER_TAP = 0b1010 /* Layer 0-15 */ }
    #define ACTION(kind, param) ((kind) << 12 | (param))

    Of course all this code is just a bunch of bit shifting!
    Let’s try OSL() one more time, just in case…
    Yeah, it’s probably because I had #define NO_ACTION_ONESHOT in my config.h, huh?
    Nope, OSL still isn’t working on the georgi. Switching to Butterstick to see if it’s a steno mode thing.
    Yes, OSL works on the butterstick. Also, implementing oneshot_layer_changed_user in keymap.c makes it print to the QMK toolbox (I wasn’t seeing those prints from Georgi).
    Switching back to Georgi for one more shot, then I’ll compromise with TG and hope I don’t get in a bad layer state.

    I got it! I had to delete the #define NO_ACTION_ONESHOT, and go into (beware there’s one for the keyboard and one for the keymap!) and set NO_TAPPING = no! Now I can use OSL to shift layers.

  • maybe write a custom function like TH(kc1, kc2), where if you tap it sends kc1, and if you hold it sends kc2. For inspiration:

    LT(layer, kc) - momentarily activates layer when held, and sends kc when tapped. Only supports layers 0-15.

    // L-ayer, T-ap - 256 keycode max, 16 layer max
    #define LT(layer, kc) (QK_LAYER_TAP | (((layer)&0xF) << 8) | ((kc)&0xFF))
    // M-od, T-ap - 256 keycode max
    #define MT(mod, kc) (QK_MOD_TAP | (((mod)&0x1F) << 8) | ((kc)&0xFF))

    Since stubbing this out and then playing with my layers quite a bit above, I realized I had an extra blank key next to the media keys, so I no longer need to double up TH(KC_VOLD, KC_MUTE), but maybe it’d be a fun experiment sometime anyway. Another day. Today my nav and media layers work great and I should take some time to practice them (and to practice steno lol).


19 Nov 2020

categories: steno
tags: plover georgi qmk

Today I updated my qmk_firmware repo [now deleted] to match what Jane has on their side. I tried a PR, but that got really mucky with a changed file and a changed submodule, so I backed out, saved patches of my commits, and completely remade my fork. It turns out the changed file also hit my new fork (wtf?), but the patches applied just fine.

Last night I read about dictionaries, and how Learn Plover! teaches the shift/ctrl/alt/super keys. It’s interesting, and I’m ready to start tinkering with my own dictionary for nav & control. I also tried using the Georgi’s pre-baked Symbol layer, but had some trouble with the alt-tab behavior I expect [see previous post].
It’s worth re-flashing with the updated qmk from germ/qmk_firmware, and if that doesn’t work I can still try my own symbol map and if THAT doesn’t work perhaps I can fall back to using a steno dictionary entries to define “alt-tab once”, “alt-tab twice” etc.


13 Nov 2020

categories: steno
tags: georgi

Got a little practice in. I realized I’m not yet ready to fully drop the progress-tracking. My reasoning is that I’m doing this from at least 2 different computers still, and Typey Type doesn’t have user accounts, so if I want to track progress there I should copy the file down after every session.
Thankfully I was able to quickly build a CLI tool that will help me merge two JSON files into one, summing the values along the way. That should help in the task of saving the progress JSON at least.

I also remembered that because Georgi only has the two rows for my fingers instead of a steno machine’s standard 3, the third thumb button (on both sides) is my number bar modifier.
I wonder again how I might use the extra two pinky keys on the left side too, since I probably won’t need–wait…i’m not so terrible at georgi qwerty! [that last phrase only took several minutes to puzzle out, and even then i didn’t know all the punctuation keys]. Anyway, even while I’m doing steno I’ll want a way to hit arbitrary shortcuts. That’ll be interesting.


20 Feb 2020

categories: steno
tags: georgi

1 day of practice in February. 13wpm on Georgi.


16 Dec 2019

categories: steno
tags: georgi

1 day of practice in December. Still hovering around 10 wpm on Georgi (using Typey Type). I started punctuation though!


6 Oct 2019

categories: steno
tags: soft/hruf georgi

That stint in August lasted 5 days. Seeing a pattern…
It’s ok though. I’m having fun with this when I have fun. Any practical effects will be a nice bonus, but not worth beating myself up about if I don’t achieve.

In August I got up to about 16 wpm on Typey Type with the SOFT/HRUF.
Now I’m using the Georgi, around 13 wpm on Typey Type.