musicmatze

Same here, atuin is such a time-saver, especially when it comes "oh dang, I ran this command on another host but I need it here now". It just works as you'd expect!

Been using it for years, for opensource stuff and at work. No issues whatsoever.

As long as tokens are cheap

I think a business case for a "look I let an LLM rewrite a large codebase" does not exist.

Shame they use discord though.

You did not explain why the patch based process is "inferior", neither did you explain why you'd have to "work around" the process!

One very crucial point that no forge (IIRC) supports that the article missed (or I accidentially skipped it) is that email supports tree-style discussion! That is a HUGE benefit IMHO, especially for patchsets, but also for "issue" discussion!

I switched from KDE with almost what you mentioned: Workspace 1 had a fullscreen terminal with zellij, Workspace 2 had a browser, workspace 3 had two chat apps open and that was it. Bindings to switch between those. I switched to niri because it was different and more lightweight than a full plasma setup at first, but now adapted my workflow a bit. Most of the time, the individual windows I have are screen-sized still, similarly organized: 1 has development, 2 has browser(s) and occasionally my email reader, 3 has my chat apps. I open new terminal windows more frequently for just firing a few commands or starting some long-running thing that I need to look at from time to time. With KDE, I had these windows in the background, now I have them side-by-side on "1". "Alt-Tab"ing between them in retrospective feels clunky now compared to Super-hjkl'ing through the windows... YMMV of course, but I think my workflow got "lighter" because of that, no more "windows over eachother" but rather "next to eachother" ... gives a feeling of lightness to me.

I see where the author is coming from, but still this feels like a reinvention of commit trailers ("Acked-by", "Reviewed-by",...) especially for these non-blocking and "only appraising" comments. Commit trailers even have the benefit that they are recorded in the comment, making them discoverable and even allowing statistics, while "human language" comments are not that easy to do statistics on (modulo AI I guess, but that's another topic).

Exactly as you do. The thing that's different at first is that workspaces are organized vertically rather than horizontally, but I for one adapted to that really quick (coming from KDE).

Unfortunately, muscle memory cannot be changed that easily. And thanks to the git maintainers, the checkout command still exists, because apparently they know that.

It is nice to see another IPFS implementation coming up, especially in Rust - but will there be any good client libraries in Rust any time soon? Does your project involve implementing client libraries?