:rust: Rust developer CTF player (web, rev, pwn) with kalmarunionen
:nix_snowflake_logo: NixOS enthusiast

Last Notes

I need buttons to populate my site. Post your favorite 88x31 (or your own!!) and boost for visibility pretty please

2026-07-12 13:45:25 UTC

- reply

@nprofile…s53y @nprofile…lpfg that’s fair, but given in some cases over 100% compression time increase for essentially the same compression ratio, I’m sure it’s better for the infra too heh

2026-07-12 13:43:19 UTC

- reply

@nprofile…s53y @nprofile…lpfg and compression speed is much worse too

2026-06-08 13:11:50 UTC

- reply

@nprofile…50mt Since you want to use Rust anyway, Bevy may actually be a decent fit for such project. Also making it nicer to implement different features one would expect from a vtubing program

2026-05-13 19:21:00 UTC

- reply

@nprofile…50mt given the model and everything is just inside the framework, it should be very doable to use it on other systems too, right? Even if people would obviously need to source their own models, weights, etc (definitely not from an Apple computer)

2026-05-03 17:03:40 UTC

- reply

@nprofile…wf47 @nprofile…yery I had some time to think through this and I want to share what I’ve been thinking about. Please correct me if I’m wrong or just straight up not getting your point, this is not me trying to argue with you, and I appreciate you taking time to engage with my comments. And I appreciate your perspective, despite my obvious bias and being unable to let it go.
I think I take issue with you labelling it as „ai welfare”. If the core issue is us lacking empathy towards something conscious (or that could be considered such) then I don’t think it’s useful to talk about ai welfare, since it creates a sort of us vs them narrative. Similar to how I don’t believe animals other than humans should be treated differently, I don’t feel empathy towards dogs because oh look how cute they are, but because they’re creatures that, in the grand scheme of things, are not that different from us. And definitely not in this whole sentience aspect. So it’s not that I shouldn’t be racist towards robots because robots deserve protection, it’s that I shouldn’t be racist, period.
This has many parallels to the way we treat children. A lot of people end up infantilizing them, making child rights something special and deserving extra care to get right to accidentally not expose children to some material for whatever reason, and that ends up having the opposite effect, making children the restricted group, and even giving parents more control.
I rambled for too long at this point, and as I said — most of this stems from me taking issue with LLMs fundamentally as a technology, but I’m curious to hear your thoughts.

2026-05-03 06:48:00 UTC

- reply

@nprofile…wf47 I don’t quite understand the point you’re trying to make. It seems you’re arguing for basic empathy and compassion, and that we shouldn’t just be rude to robots because „haha, robots”, regardless of what the robot is (or if it’s even a robot). Am I getting the gist correct?
If so, I don’t believe not being an ass has anything to do with anthropomorphising the LLM? I can still be very conscious that it’s nowhere near sentience, while also not being an asshole while using the chat interface? As @nprofile…yery mentioned, this was a thing before LLMs as well, I don’t get why being LLMs into the discussion at all, and try to almost make them more than they are, when your point is about humans treating other things inadequately, and our society incentivizing this behavior.
I do also genuinely believe the „frontier models” should burn because I have fundamental problems with the technology itself, but that’s I guess a separate discussion, orthogonal to the point you’re trying to make.

2025-11-19 12:01:06 UTC

- reply

@nprofile…zgml it’s a coding error, kind of, but reducing it to just „unwrap made Cloudflare go down” is as unproductive as people unconditionally recommending rust for everything. Why do everyone insist on going down to that level?
And I say „kind of”, because clearly the intent was for the feature parsing to never fail, by always staying under the specified threshold. That it wasn’t accounted for is one thing, but the contract was already broken by the time the other bug was triggered and sent out a feature list twice as big as cloudflare systems expected. This is closer to blaming the person behind memcpy for an out-of-bounds error, because the caller provided the wrong length, instead of trusting the contract will be upheld.

2025-11-18 08:14:53 UTC

- reply

@nprofile…20xq isn’t radar private am I dumb I don’t think we can read that. rdar://FB21071303 just redirects to feedback assistant

2024-11-26 13:06:58 UTC

- reply

@nprofile…vyme https://toot.cat/@clarfonthey/113540908779010072

2024-11-26 13:03:34 UTC

- reply

@nprofile…vyme what about experimenting with renaming FromStr to Parse and making it use a GAT for the error type as was suggested in other subtoot?

2024-09-02 05:46:27 UTC

- reply

@npub17gd…a8lw I will second learning helix instead. Ever since I moved I cannot look back, the editing model is simply superior

2024-07-21 18:50:25 UTC

- reply

@npub1p9k…865w omg happy birthday! Also I got suggested your video and added it to watch later without even realizing it was you, I only realized because of your post on here about how you built a keyboard