Hacker News new | ask | show | jobs
by SSLy 1851 days ago
But we shouldn't held UX ransom by such hardware. Martix has TUI clients too.
2 comments

Currently playing with Matrix a bit and I have to say that most of the TUI clients are pretty terrible still. The desktop clients are not much better, they either have mobile UI but on desktop or are quite buggy still. And what they all have in common is that they are quite slow as well, despite local homeserver running off an nvme ssd.

None of the ones I have tried come even close to UX of pretty much any desktop IRC client.

weechat-matrix should be pretty great, given the TUI is... well, weechat, which is a pretty great IRC TUI?
That had some weird interactions with some of the other scripts I was running. There were some known incompatibilities that were still being worked on - and now it's being rewritten in rust? In any case, I wasn't having a great time with the python version of it.

So went with a more native TUI application (gomuks) and didn't really enjoy that either (scrolling broken, weird mouse behavior, unread messages got stuck on some channels and in other rooms it didn't show any unread messages at all, stuck room that I already left and some other issues that annoyed me).

Then I thought I give some desktop applications a go (quaternion, spectral, nheko, mirage and last I tried was neochat) and they fell into roughly two categories: ok-ish UX but too buggy to be usable or work but have UI designed for mobile but on desktop (which is a no-go for me).

That's where I am with my quest for a decent matrix client. I haven't tried _all_ desktop clients yet, but there are not that many non-electron/browser based ones left I think.

>which is a pretty great IRC TUI?

Catgirl, or some patched mpu irc.

It really doesn't. The best option is weechat-matrix, but that's already a dead project. The lack of good client diversity is what is holding me back from using Matrix.
Dead is overstating it. The maintainers are moving new feature development to a Rust rewrite, weechat-matrix-rs, but still accepting bugfixes/etc to the current Python codebase
Hmm. Given we got here by talking about people who can't (rather than won't) run heavyweight GUI clients, a Rust rewrite seems like it misses the point.

That is, I'm sure it makes sense for them, and for most of their actual users, because Rust is nice, but if I have an m68k machine, too bad, there is no Rust for m68k (unless I'm reading the platform support chart wrong) and so this rewrite likely orphans me.

Now, if for a Debian channel, even a Debian channel about an attempt to port Debian to m68k, then I have no doubt your users all have something less stupid, maybe ARM or x86-64 or even PowerPC. They prefer a non-GUI for their own reasons but ultimately Matrix is an option.

But if you're a channel for actual m68k "classic" Amiga you might actually have a non-trivial number of users for whom "it doesn't run on m68k" is a showstopper. They have an IRC client today.

Rust doesn't target barely-capable archaic platforms like Motorola 68xxx or 80286 and I don't think it should start. But that means we have to accept that a Rust client won't support those platforms.

I'm not seeing classic Amiga on Python's supported platforms either: https://www.python.org/download/other/

I doubt it would perform well even if you did get it running.

So this isn't a change in platform support.

Fair point.
weechat-matrix is really not a dead project. the maintainer works for Element, and is working on matrix-rust-sdk (and thus weechat-matrix-rs), but weechat-matrix is alive in well and in no way unmaintained or dead.
Okay, but I'm not going to go through the effort of setting up weechat-matrix if I know I'm going to need to later throw all that effort away and switch to weechat-matrix-rs. I'll just wait for the latter. It is, for my purposes, dead.