>Let’s cut short: I’m really interested to be one of your customer (as>long as you are in my budget and you can deliver to Europe in a way or>another). That’s really the kind of service I was hoping to see: a>custom computer maker. That’s awesome. I understand that this would, of>course, not be a cheap computer.
It would truly be an honor to work on something for you, and delivery
is possible even in the case that I have to fly it out myself. I'll
follow up directly to start chatting about details.
>The usync/dsync is really a great idea. I guess those scripts could be>extended to allow synchronisation through USB with a remote data>storage.
Absolutely, this is one of the benefits of having a shell/kernel — you
can think up little utilities for nearly anything and implement them
locally and quickly. True programming, I would go so far as to say.
>I guess we can easily include tools that are now mostly standard:>python, ripgrep, fdfind, less, pandoc, git and a few others.>Are you running X or Wayland? (if yes, a WM like i3 would make sense).>Another question: does it have audio? This is something I’ve mixed>feelings. I don’t know if it is a good idea or not.>For my part, I would really be interested by an ethernet connection to>synchronize Offpunk and my emails (currently neomutt/msmtp/offlineimap).>It would be awesome to have a physical button that would :>1. enable ethernet if it is plugged.>2. Run do_the internet.sh (my infamous script which sync offpunk/git>repositories/email)>3. run usync/dsync if an USB is plugged.
Adding other software tools is very easy, just a few lines of config.
Same for a WM like i3. NixOS is pleasantly mature these days. I could
certainly physically expose the ethernet jack and add another tactile
button (functionally a hotkey) with a small software interface in case
you'd like to bind it to another command as well.
>What’s your opinion of MNT?
I really like MNT as a company/research org, they are thinking about
the right sorts of things and anyone doing good work in this space is
a very good thing. The pocket reform and reform are interesting ideas,
but they don't strike me as an order of magnitude better than the NUC
I used. It's very nice that they've OS'd the hardware (a wildly rare
thing to do), but drivers and part compatibility are big pains in
hardware development — when you have a system that incorporates some
funky components, adding stuff on top gets funky. It's faster and
nicer to just use stuff that you know will work and work well. So
overall: although a great first stab at an OS computer, it is
worrisome in its lack of maturity and widespread adoption (as opposed
to the incumbents).
>That was one of my key requirement: I would like a computer with an eink>screen. It really change completely the way I interact with a screen>but, I agree, a lot of the software stack should be completely>rethinked. Take the terminal: instead of scrolling line after line once>you reach the end of the screen, it should instead refresh the screen>and bring you immediately to the the top (kind of a "clear" command).>Also wondering how to use Vim with eink (the Freewrite managed to do it>so it is not impossible). Another alternative would be LCD screen.
Performant eink is by no means impossible (see e.g. the Remarkable
tablets https://remarkable.com/), but it is time-intensive for the
reasons I mentioned. Most products that have been able to achieve what
I would consider 'acceptable' performance are produced by large
engineering orgs, so for one guy it's tough. They also decay
irreparably with use over a period of a few years (I believe ~3 years
was a number I saw somewhere), and they decay faster the faster you
push them (i.e. bumping refresh rates). You mentioned an LCD as an
alternative — the Mythic I does indeed use an LCD. There are
monochrome LCDs as well that are much easier to interface with and
have better longevity than einks, so I think they're not a bad option
either. In general I lean towards more established technologies
because they are easier to interface with and tend to be more
reliable. Luckily the wood will last for centuries in nearly any case.
>Another pet requirement of mine is the keyboard. A good, mechanical,>orthogonal keyboard. Examples:
I love the moonlander — it could certainly be used! A design
consideration though is that for each cm of width you add overall, you
add quite a bit of weight.
-Keegan McNamara
On 23/03/30 03:38, Keegan McNamara - keemcn1 at gmail.com wrote:
>>>What’s your opinion of MNT?>>I really like MNT as a company/research org, they are thinking about>the right sorts of things and anyone doing good work in this space is>a very good thing. The pocket reform and reform are interesting ideas,>but they don't strike me as an order of magnitude better than the NUC>I used. It's very nice that they've OS'd the hardware (a wildly rare>thing to do), but drivers and part compatibility are big pains in>hardware development — when you have a system that incorporates some>funky components, adding stuff on top gets funky. It's faster and>nicer to just use stuff that you know will work and work well. So>overall: although a great first stab at an OS computer, it is>worrisome in its lack of maturity and widespread adoption (as opposed>to the incumbents).>>>That was one of my key requirement: I would like a computer with an eink>>screen. It really change completely the way I interact with a screen>>but, I agree, a lot of the software stack should be completely>>rethinked. Take the terminal: instead of scrolling line after line once>>you reach the end of the screen, it should instead refresh the screen>>and bring you immediately to the the top (kind of a "clear" command).>>Also wondering how to use Vim with eink (the Freewrite managed to do it>>so it is not impossible). Another alternative would be LCD screen.>>Performant eink is by no means impossible (see e.g. the Remarkable>tablets https://remarkable.com/), but it is time-intensive for the>reasons I mentioned. Most products that have been able to achieve what>I would consider 'acceptable' performance are produced by large>engineering orgs, so for one guy it's tough. They also decay>irreparably with use over a period of a few years (I believe ~3 years>was a number I saw somewhere), and they decay faster the faster you>push them (i.e. bumping refresh rates). You mentioned an LCD as an>alternative — the Mythic I does indeed use an LCD. There are>monochrome LCDs as well that are much easier to interface with and>have better longevity than einks, so I think they're not a bad option>either. In general I lean towards more established technologies>because they are easier to interface with and tend to be more>reliable. Luckily the wood will last for centuries in nearly any case.
Another idea I had was to experiment with a screenless setup : directly
printing on paper. The setup should, of course, be tested but I was
wondering how manageable it was to write by printing directly, using
potentially ed and less to edit and read file.
That could be really interesting. I don’t know if anybody on this list
has an experience with that.
On Fri, Mar 31, 2023 at 09:24:43AM +0000, Ploum wrote:
> > Another idea I had was to experiment with a screenless setup : directly> printing on paper. The setup should, of course, be tested but I was> wondering how manageable it was to write by printing directly, using> potentially ed and less to edit and read file.> > That could be really interesting. I don’t know if anybody on this list> has an experience with that.
What I would suggest in that regard is ed and espeakup so you don’t use the paper resource. I have used ed every day for all editing tasks, both journaling, coding, and blog composition.
I have used espeakup a certain amount with screenless computers, e.g. old laptops with broken screens and found it mildly usable but definitely needing refinement.
Towards that end I wrote a z script (like the z command in my ed which prints a screen of lines) which I think of as the “easy” command. The z script runs any command but saves stdout and stderr and simply replies with success or failure. Then I can go back and use ed to look through the log. This reduces the amount of speech produced by espeakup.
My ultimate goal is to make a system which would only need one button for more code input and a microphone for simple local speech/sound recognition instead of a full keyboard.
Also, in thinking about the forever computer, I have thought that the conclusion would be that myths/stories and human connection is the real forever computer. Think about how many people have "retained" various traditional stories and knowledge and such!
Love this list. Be well!
Craig