For comparison, I've used PostmarketOS on Pinephone, and it required a lot of fiddly to get a poor experience.
But maybe the simpler usecase of "just reading" has good solutions?
For comparison, I've used PostmarketOS on Pinephone, and it required a lot of fiddly to get a poor experience.
But maybe the simpler usecase of "just reading" has good solutions?
I definitely stay with my bank (Lloyds in the UK) partly because they have a good website, and I will not bank with HSBC because their app will not work if you install things from outside the Google App store (and logging into the website needs the app, at least for me at the moment - I think that can be solved).
I have this requirement too, since I like to use F-Droid.
My point isn't that there are no such users. My point is that product managers in banks don't care about F-Droid users, since there's so few of us that it's not worth them worrying about.
Many websites are giving up Firefox support, and Firefox adoption is much higher than F-Droid.
If a bank app happens to be okay with F-Droid, it's not because they look out for the needs of F-Droid, it's simply by happenstance.
But I'm not sure about his other stuff.
"Avoid features that add disproportionate cost"
I expect part of the problem here is that it's often not clear what the value of features until it's available to customers.
Even the costs of bloat are unclear. Take his bank website example. Do we really think many bank customers are choosing banks based on their website's latency? Banks compete on things users actually care about, like interest rates or fees.
Lots of software inevitably won't meet our ideal standards, because given the cost of developers it's not worth doing things The Right Way.
Window managers can plausibly already do a lot of what other software can do, yet in practice, popular workflows tend to assume very little from the window manager.
I try to avoid terminal multiplexers in favour of Sway/Emacs/dtach/SSH multiplexing, but I still often reach for tmux.
Source: this is a description of my own workflow and preferences, so I’m the ultimate authority on the subject, haha.
It's fine to choose your workflow by whatever criteria you decide, but on a post about workflows on a discussion forum, it's reasonable for mvdtnz to continue that discussion and not be laughed at for doing so.
It's nice to have to a set of programs I moderately care about, so when I'm learning another language I can port them to it.
They're unimportant enough that I can comfortably experiment, but important enough that I want to complete the rewrite.
My "blog engine" is a nushell script that uses pandoc and built-in XML support to convert markdown into a site+feed.
https://discourse.nixos.org/t/how-minimal-can-a-nixos-image-...
NixOS works well for x86-64 and aarch64, but not so much armv7l, as so many consumer routers are.
The PC Engines happens to be x86-64 with decent storage expansion, but for sure if you want to target armv7l, NixOS is not a good choice.
I see the fresh re-install suggestions probably work but that's tedious and risky.
That's a misconception. Nobody actually cares about security for packages that are not in the default install. For example, the initscript for sstp-client disables certificate validation unconditionally, see https://github.com/openwrt/packages/issues/25212
> Nobody actually cares about security for packages that are not in the default install.
Probably an exaggeration, but it's clear there are some packages that are insecure out-the-box.
For comparison, I've used PostmarketOS on Pinephone, and it required a lot of fiddly to get a poor experience.
But maybe the simpler usecase of "just reading" has good solutions?