Sven Slootweg (🔜 WHY2025) @joepie91@pixie.town

@eniko WASM ran into the same problem but worse with "every binary shipping its own GC" and I believe that this is why there's on-going design work on a sort of shared GC API that isn't necessarily a whole GC, but provides a bunch of mechanisms to share one across different WASM 'processes' without breaking sandboxing.

I forgot the details but perhaps there is something of interest in there for the issue you're having?

@MrBerard@pilote.me That got a bit longer than I expected 😅 Hope that's helpful? Feel free to ask about any of it!

@MrBerard@pilote.me Partly - there were a few different reasons though they're all kind of related:
- I find institutional education, with a rare few exceptions, to be a hostile and harmful environment to most people in a way that is fundamental to its model (despite the best efforts of many teachers!)
- The expectations that those environments would have of me as a teacher, also would not actually be satisfiable for me as an ND person
- I have a rather specific teaching method that to my knowledge no institutional setting uses, and for which there would also be no room in such a setting (except for *maybe* a few special education places)
- The scale of most institutional education is completely unsustainable, in part because of the hierarchical teaching model used (with a strict division between teacher and student roles)

I have many more thoughts here, but those would be the high-level summaries - what it boils down to is that the way in which I can teach most effectively (and which I know works; I've not had a single student that didn't get there) is not something that there seems to be any room for in institutional education, where from what I've seen, procedure is usually valued over adaptation.

I'll ramble off a bit here about the way that I teach software development, hopefully it'll illustrate why I've not found a way to fit this into an institutional model:

It's a heavily personalized project-driven approach, where the student selects a project idea (free choice, of arbitrary complexity, even if pie-in-the-sky), and I help them learn how to divide that idea down into smaller components and smaller gradual steps to get there, starting with the absolute simplest (and functionally useless) facsimile of the thing they want to make, and then gradually building on that.

Then we go through the first steps together, for an hour or two, and by that point I usually have a pretty good idea of what their existing skillset is like, and what things they tend to have trouble figuring out. I explain some of the basic concepts and approaches, as well as how to figure out issues with them by themselves.

From that point on we switch to a regular schedule where there's usually one or two sessions a week, 2-3 hours each, to talk through the things that they got stuck on. The rest of the week, they work independently on their project, usually working on a few different bits and pieces in parallel - if they get stuck on one, they table it to discuss it in the next session, and continue on another in the meantime.

Throughout the sessions I ask them to explain what they got stuck on, and then don't give them the answer directly, but guide them through the process of *finding* the answer, asking steering questions until they arrive at the answer by themselves and providing hints (usually in the form of questions again) where needed, and then usually doing a round of "okay, so can you now fully reason to how you got to this answer and why it is the right one?" - the goal being to help them build confidence in their ability to find the answer themselves, because that is often the actual thing blocking them.

In the end, this process doesn't actually have much to do with conveying knowledge; it's sometimes important to steer someone to the right path before they get frustrated, of course, but I consider it far more important to teach the 'meta-skill' of figuring out unknown things, and building the confidence to try and explore possible solutions to problems, "teach them to fish" so to say.

All of this is... very different from anything I've seen in an institutional setting 🙂 Which tend to be heavily focused around rote memorization, somehow magically expecting "things that were said" to convert into "things that are remembered", and where testing procedures don't actually test for anything that's relevant to people's lives... with perhaps the exception of (a subset of) vocational schools, which have their own problems that make this model not really viable.

@miyuko I guess to phrase it differently: Preact used to come across to me as a very... "toxic minimalism" project, whereas nowadays it seems much closer to the "React but without genuinely unnecessary complexity" that I actually wanted

@piegames Unfortunately turning off boosts doesn't really work because a lot of interesting stuff ends up in my timeline through boosts... filter rules are a bit annoying because I'm stuck on Mastodon 3 for now and it doesn't let me auto-CW things, only hide them entirely 😐

(News channel aren't really an issue, almost anything with news channels is instance-blocked here and the remaining few I muted long ago 🙃)

@miyuko It doing hooks now is a big one, but last time I looked there also didn't seem to be any (well-documented, at least) support for server-side rendering (which is quite important to me - I use React as a server-side templater primarily!), and IIRC it used to have a much stronger "absolute smallest possible subset of features" philosophy, whereas now it seems to be a more reasonable "no unnecessary things but also not unnecessarily minimal".

Signals were a nice bonus that I only found out about later :)

@roberth @Janik@chaos.social Oops, to clarify, when I say "happening in Nix", I mean in the broader community. This is the kind of thing that requires pretty wide community support to have any hope of dealing with in this manner, and I don't think this broad support base exists in Nix.

@roberth @Janik@chaos.social I don't know. I've seen this play up close in a *lot* of FOSS communities, including the same "make sure the community understands that upstream is what matters", and I have literally never seen this work, ever. Anywhere.

The only thing that has worked reliably and without significant collateral damage, has been a governance-level solution - ranging from outright banning proprietary companies, to in milder cases deliberately de-emphasizing them and setting extremely strict rules about how they can participate, to in a rare case setting up internal competition through multi-corp governance (but Nix is probably not in a position to make that last one work for various reasons).

None of these things are happening in Nix, not even anything that looks remotely similar. The outcome of this path is that the effort to protect upstream will fail, and in 10 years there will be retrospectives asking "how could this have happened?"