Notices by Henri (slyecho@mdon.ee)

@kaia Everything other than Apple devices on Apple computers will sound like a telephone in my experience.

@dansup I think storage is still going to be a limitation, when talking about video. Better codecs? Peer-to-peer something?

@eniko Are you on Windows? It's usually much harder there. But generally, using the CMake extension and making sure that also works will allow the C/C++ extension to find all the stuff. Also, depends on what compiler you're using.

@eniko people here complain about Bluesky, on Bluesky people complain about Twitter and make fun of Mastodon, on Twitter they complain about Twitter and do all the previous, on Threads they make fun of Apple Intelligence.

@hyc @dalias @ariadne Maybe they could unfold like the panels on satellites and stick 30m up in the air. Wind could be a problem, though

@lanodan It's been sometimes useful to have the old package to downgrade on some unstable updates.

Otherwise pacman -Sc

@lanodan @kura You can do dynamic linking also with -rpath $ORIGIN etc. and then release just binary tarballs with executables and libraries together, kind of like on Windows.

@dansup @loops What aspect ratios will you have? This one seems like 3:4 vertical?

@lanodan Well, I guess we are in agreement then that the default char limits on Mastodon are too small.

@lanodan @novenary Mastodon itself and other apps will display the messages fine but there may be some "read more" button. It's also possible to have HTML formatting in posts as well, even if Mastodon itself does not enable adding it in its editor.

You can check on some posts by @marcan for example for how it looks.

@lanodan I depends on the server, I have a 10k limit.

@dalias @chris @realhackhistory There are no error codes, the model is trained to respond like that, or in a similar wording because it's a language model after all.

There are actually out-of-band signalling tokens that are not mapped to Unicode input or output, but the model can also ignore them just like a typo in the user's text.

@dalias @chris @realhackhistory If you have the possibility to run your own algorithm to select the next token you can do much more powerful things, like constrain it to a specified grammar, etc. but OpenAI does not offer that interface to users.

@ariadne Actually having errors as a language feature would be better even, like Zig or Rust