It was a matter of many months, although not 100% all the time on it. I started with trying different designs for a single "down" key mechanism, then just kept broadening the scope - creating a single "cluster" of 5 keys and doing a little proof of concept with the electronics on it. Figuring out the thumb mechanisms.. figuring out the handrest, and the adjustable cluster mounting mechanisms, etc. I wouldn't say there was any particular "failed approach" per say, just a lot of little failed design details along the way, that were changed/improved over the course of the design.
It would probably take me about 40 hours (of me time - not including 3d printing, etc.) to make one from scratch at this point. For someone trying to figure it out for the first time, I would double or triple that at least. But once I finish some design tweaks I've been working on, which include some actual fabbed pcbs, it shouldn't be as bad.
I'm honestly not convinced it's actually copper filled. The listing says it is, but its density says otherwise. Regardless, it was the most opaque thing I had on hand, so.. :)
There's some good info on opaque filament that people have found on this thingiverse object: https://www.thingiverse.com/thing:1662956. Personally, I've also had good luck with eSun's black PLA, although it was a fairly old roll, so who knows if their formulation is still the same. And their black PETG definitely isn't very opaque.
The vinyl cut thing was super useful during the prototype stage, while things were still in flux, and the much shorter turnaround time was advantageous. Although it turned out to be a little side-project all its own, so not entirely sure if it was worth it. But it was a fun side-project none-the-less. And that's what counts in my book, at least!
I guess it's a personality thing. I actually love that whole experience - both when I switched to dvorak on a normal keyboard, and then again when I switched to a datahand.
I love the process of developing muscle memory, the mindless repetition, and that feeling of improvement. And especially once it really starts to click, and your fingers start going on autopilot.
If you want something to make it a bit less of a drudge, you can take a look at the Amphetype software, which lets you retype ebooks -- so you can basically read a book while practicing typing it.
It does technically require chording, but honestly not much more than a normal keyboard. Just a few extra "shift" keys, more or less. And normal textual input is pretty much all non-chorded.
I'm a tiny bit biased, of course, but I do find it quite usable. Even (or especially!) for gaming. But yeah, I've been using a datahand, and then later my lalboard as my primary keyboard at home and work for probably going on 15 years now.
I actually prefer number entry on the datahand. They're such a reach from home position on a normal keyboard, but they're just as easy to hit as anything else, with a datahand-like.
[4] That's me! :) It is quite comfortable to type on. And it's actually great for mouse+kb gaming too. I have 20 keys right at my fingertips, and I typically map all 3 modifiers to the thumb. Although I do recommend more of a straight across asdf (backwards/left/right/forward) movement scheme (or aoeu for those of us of the dvorak persuasion).
The 120WPM was a bit of a cherry-picked example. 95-100WPM is more of a typical average speed for me.
Yeah, most of the heavily used keys are mapped to north, south and down. There are a few letters mapped to side keys, but they're mostly on the first two fingers. The first finger can do either side direction fairly easily, while the middle finger can do the inward (toward the thumb) direction a bit more easily than the outer direction.
The smali code that is generated is in fact reassembleable. The smali format is meant to be as close to a 1-to-1 mapping for the dex file as possible. Disassembling and then immediately reassembling something will produce an equivalent (almost identical) dex file. The only things that might be different are misc. internal details, like the order of certain items, etc.
It would probably take me about 40 hours (of me time - not including 3d printing, etc.) to make one from scratch at this point. For someone trying to figure it out for the first time, I would double or triple that at least. But once I finish some design tweaks I've been working on, which include some actual fabbed pcbs, it shouldn't be as bad.