Replying to
@eniko@mastodon.gamedev.place
all girl dungeon town where you get to date the cute shopkeepers
Seeing the crt shader applied to it from the other post I was entirely expecting a game with intense ambience and horror lol
Remote
λzyd
@zyd@yap.zyd.lol
Posts minus the boosts: @zyd@yap.zyd.lol
0 Followers
0 Following
50 Posts
Joined November 10, 2025
writes:
uses:
lisp, emacs, and other miserable little programs
enjoys:
anime, manga, computers, strange tunes and whatever else
Replying to
@rowb1t@social.xenofem.me Are y'all doing any of the exercises? Curious about the format or if it's more laid back
Replying to
@dwardoric @shinmera @valorzard Yeah, it's brutal since SBCL is the best performing Lisp compiler. I still intend to use SBCL but with my fork of it:
https://codeberg.org/zyd/sbcl
It's based on the recent 2.6.2 release with the leftover bits of slop removed, so it's a relatively clean slate. My hope (perhaps naive) is that it provides a good starting point for a future hard fork, should some sufficiently motivated and principled compiler hacker(s) want to take SBCL down a different path.
Replying to
@shinmera @valorzard Slightly related, a little bit ago I decided to be annoying and ask nearly every major Lisp/Scheme for an AI policy/stance:
https://zyd.lol/lisp-against-the-machine.html
The Lisp story with LLMs is unsurprisingly not so good.
Replying to
@screwlisp Yeah, for context: mine exists specifically to address the friction of learning Common Lisp (or Coalton). It's a single executable that bundles a compiler (SBCL), a Terminal and Quicklisp, all into one so any noob gets a double click->Lisping setup. It's for noobs, not those already Lisping and know Emacs. So it does things like CUA bindings, non-extensible etc. The idea is that folks can graduate to Emacs later on if they want (though I suspect if they start off with mine, they'll never learn Emacs but shrug).
Though I've long been one of those that doesn't think learning Emacs (along with Quicklisp and ASDF) is asking too much, testing mine in the initial alpha group kind of convinced me that yeah, this probably should have existed a long time ago. Emacs and other associated setup can no longer be used as an excuse to not learn Lisp.
@tux0r @svetlyak40wt @vindarel @sanityinc @simon_brooke @jackdaniel @dougmerritt
Replying to
@dthompson@toot.cat Or maybe not thank god if its gonna be taken over my Emacs LLM enthusiasts (I already see one of them offering up)
Replying to
@dthompson@toot.cat Thank god as I discovered just the other day he fell into the slophole and was wondering if elfeed was at risk.
Replying to
@neauoire I hope one day humanity will have advanced enough that we can dedicate an entire museum to cursed units of measurement. Each unit should have some kind of story, perhaps a tale of great warning!, to go along with it.
Replying to
@mirabilos@toot.mirbsd.org @lisyarus@mastodon.gamedev.place Do the tags not tease enough that it's probably not about snails? Or how about the fact that the person posting is usually talking about graphics, gamedev, and programming. You could click the article and spend about 10 seconds deciding if it's interesting or not. Talking about needing to be enticed sounds like you have too much of a firehose of a feed going on.
Replying to
@dillo You could also use sourcehut's paste service: https://paste.sr.ht - But you'll need a free account. I upload pastes via the hut command line tool for it with hut paste create --name whatever-title-i-want file-to-upload.txt
Sourcehut works well without js and from the command line.
Replying to
@jk why do you have to be polite to mastodon's ui lmao. "No thanks" as if I'm talking to front desk.
Replying to
@technomancy Have you used git-pr before to do collaboration or know anyone that has? Could use a "user experience report" type thing.
Replying to
@demha@comp.lain.la
Feel that. My primary opposition to LLMs isn't even based on ethical grounds, but socioeconomic. It's a technology made specifically to eliminate large amounts of labor, and intensify the remaining. It's a direct means of overexploitation. And it's hard to say how applicable sustainability is for these monopolies given how much they're embedding themselves in state bureaucracy, at least in the U.S.
There's also just the basic fact that capitalists have had a taste of a tool that, regardless of how shit the result is, can eliminate large amounts of workers. They aren't going to give that up. I have nothing optimistic to say unfortunately.
@evgandr@mastodon.bsd.cafe @valorzard@mastodon.gamedev.place @shinmera@mastodon.tymoon.eu
Replying to
@hacknorris
Yeah, that's what I was doing as well: using tabs as a poor man's bookmark management tool, because Firefox on its own doesn't care about you actually using bookmarks to any real degree. I basically never had many of those 400+ tabs actually loaded.
Just in case you might want something different, you might like linkhut for trying to deal with this instead. It's something I've recently started heavily using myself along with the extension linkput making it extremely convenient to bookmark all the things. It's open source, bookmarks can be imported and exported via a single HTML file via the web or through various means on the command line (or other external tools) through its 1:1 Pinboard API. Along with other people using the flagship instance of it, it also turns into a great search tool in general, not only for just your own bookmarks. Also has RSS feeds for any combination of tags and user profiles. It's sick as hell.
Replying to
@technomancy I mean, I feel ya, but the only solution is either an extension like Stylus or writing some kind of hacky userscript userchrome thingy. At least Stylus isn't something you need to be updating much.
I'm pretty much past the point of expecting Firefox to do anything that supports users being in control of the browser. We both know that's long sailed my friend.
Replying to
Replying to
Replying to
@deech@mastodon.social I've learned that the stated contribution policy isn't the full story: https://github.com/jank-lang/jank/discussions/758
TLDR:
This policy doesn't prevent core jank developers from using LLMs, since they are trusted to do so wisely. This policy also doesn't prevent LLMs from being used for reviewing code.
Replying to
@rpluim@mastodon.social Dunno but laws can change and they usually benefit holders of capital. Being opposed to LLMs merely on copyright grounds is a weak position I wouldn't want to be standing from. Any strong opposition has to be based primarily on the social element of LLMs.
Replying to
@technomancy Side note: If you uncheck the "allow web pages to choose their own fonts..." than this applies to all web pages, not just unstyled ones.
But you could do this with Stylus to write a stylesheet that applies to everything. Like attached screenshot does what you want for danluu (of course it'd be better to write the Stylesheet to only apply to his domain, that's usually what I do)
Replying to
@screwlisp
You can see here the definition of the math package, just as an example:
https://github.com/coalton-lang/coalton/blob/main/library/math/package.lisp
Useful reference that lists all the packages of interest:
https://coalton-lang.github.io/reference/
Also see the examples dir in the coalton repo.
@tux0r @svetlyak40wt @vindarel @sanityinc @simon_brooke @jackdaniel @dougmerritt
Replying to
@lizzy why's that? (if you've talked about this elsewhere and don't wanna repeat yourself i don't mind just a link to something, genuinely curious)
Replying to
@andreamonaco @shinmera @valorzard GNU projects are weird in that they seem to operate as largely independent sub-projects in a lot of cases. So, in the case of CLISP they've already adopted a strange LLM policy of essentially no more than 5 lines per LLM prompt. It's based on a bizarre reading of copyright law and is pretty weak.
See:
https://gitlab.com/gnu-clisp/clisp/-/work_items/62https://gitweb.git.savannah.gnu.org/gitweb/?p=gnulib.git;a=blob;f=HACKING;h=72c9895c0f407a28f5c5c82b0173ccf6e89b4429;hb=HEAD#l226
So I won't assume that because something is a GNU project they won't accept LLM contributions despite that being the case for a lot of GNU projects (e.g Emacs). And I definitely won't be listing Guile as even "Weakly Against" as I've heard nothing back from anyone on the Guile team despite leaving both a ticket on the repo and sending a message out to the mailing list. I suspect they're not saying anything until GNU lays down official policy (which more than likely will be something along the lines as LLMs are acceptable if the model, model data, etc are all local and open source; meh)
Replying to
@deech @shinmera @valorzard That's a recent change I didn't see, good lookin. They used to accept. (See the commit history of contributing file)
Replying to
@sigrid @dillo Took a couple minutes but figured out how to use it with curl, but sadly the server doesn't return the filename of your upload :( - Do you run the service or know who does?
curl 'https://okturing.com/' -H 'Content-Type: application/x-www-form-urlencoded' -X POST --data-raw 'fake=fake&a_func=add_post&submit=submit' --data-urlencode a_body@file-to-string.el
Think I managed to rip out all the remaining slop from SBCL's 2.6.2 release. Builds on Linux x86-64 so seems fine for me™. Probably going to use this personal fork for the foreseeable future. Might try to update it to future releases but depends on the complexity and amount of slop.
(Stainless) Steel Bank Common Lisp: https://codeberg.org/zyd/sbcl
The name is very cute, admit it.
Replying to
@technomancy Gotcha. I think Simon Tatham's approach is more applicable to me anyway (and resonates) but was curious. Would like to see more options to contribute to projects other than the Github workflow. Though I use sourcehut, can't say I exactly like or prefer email-based patches, it's just the only existing alternative that lets me not use a browser (I like to reduce context switches when they aren't meaningfully changing what task I'm working on).
https://clojure.org/dev/contributor_agreement#_no_generated_code 👀
Clojure’s code is written and reviewed by humans. Code generated by a large language model or similar technology, such as Anthropic’s Claude, GitHub/Microsoft’s Copilot, OpenAI’s ChatGPT, Facebook/Meta’s Code Llama et al, is not compliant with the covenants and representations of Clojure’s Contributor’s Agreement, and is thus not acceptable as code for Clojure.
I'm aware of Hickey's sarcastically titled post "Thanks AI!" post but didn't expect a formal position tbh. Good stuff to see.
Replying to
@jackdaniel@functional.cafe sbcl has had a lot of releases since 2.4.4 (current is 2.6.5), wonder if more recent will change those numbers.
Replying to
@evgandr Strictly against means without exception, until the end of time, the project is against LLMs in any sort of way. I created a category below that, Strongly, because in the case of Loko Scheme, the author is against LLMs but sort of leaves the door open to them potentially being useful at some point in time:
That said, LLMs can be useful. They can search the literature for ideas, suggest implementation strategies, help with bug hunting, double check your code, etc. Maybe they could even write test suites, but I wouldn't want them committed into the project itself.
Plus, if you open the issue you'll see that while they're against LLM generated code being included in the project, there isn't a stance against LLMs from any social/ethical grounds. You can really only be Strictly against LLMs on grounds other than technical or copyright.
That's how I interpret it.
@shinmera @valorzard
Replying to
@MURRUMUR@social.xenofem.me nice to see herbie hancock and lil ugly mane on there
wonder if you'd like this album, one of my favorite jazz albums of all time: Eddie Gale - Ghetto Music
https://youtu.be/G8nB5gaQggs
Eddie Gale (Usa, 1968) - Eddie Gale`s Ghetto Music (Full Album)
Lisp propaganda from 1988:
"Every time a program change is necessary it means starting the cycle over again."
Really like these articles' visualizations of transforms, especially the model→view→projection. Other articles do a better job at in-depth explanation (e.g LearnOpenGL chapters) but that's fine since these seem more like notes than anything.
https://jsantell.com/3d-projection/
https://jsantell.com/model-view-projection/
Replying to
Replying to
@hacknorris @kenney Oh my god, this screenshot immediately overwhelmed me. You have so much going on. This reminds me of the days when my ADHD unmedicated ass unironically had over 400 firefox tabs.
Replying to
@lizzy Do you feel the same extends to projects like Guix and Nix? I bring it up because while they're distros in their own right, they can also sit on top of one, much like Flatpak et al. I guess this boils down to why someone installs a particular distro (currently a 5+ year Fedora user eyeing Gentoo, I like a relatively stable system but with packages that don't remind me of Debian).
I asked emacs-devel: Are LLM contributions welcomed in GNU Emacs by the maintainers and contributors?
Eli Zaretskii:
We are awaiting the decision by the GNU Project on these matters, which will define the policy for all the GNU packages, and in the meantime we don't accept LLM-generated code, as a precaution.
For now, Emacs won't accept LLM contributions. However, given this:
Obviously, the right thing to do is protect computing freedom: share complete training inputs with every user of the LLM, together with the complete model, training configuration settings, and the accompanying software source code. Therefore, we urge Anthropic and other LLM developers that train models using huge datasets downloaded from the Internet to provide these LLMs to their users in freedom.
I sort of expect FSF, and therefore GNU along with Emacs, to be somewhat accepting of LLMs eventually, if they are "open". Too copyright-brained for their own good. Unfortunate.