@leah thanks, this reference helped me understand Rust's scoping/hygiene rules for macros a lot better!

Unfortunately this requires all the other macros to be #[macro_export]ed as well. bleh. I can live with it I guess.

A bit disappointed that this $crate stuff is necessary at all. Do you know if there's a reason for this "mixed-site hygiene" instead of Scheme-like full hygiene?

I was told Rust's declarative macros had hygiene. I was lied to. I seriously can't export a declarative macro that uses other un-exported declarative macros and have it work? wtf?

DumPy: Like NumPy except it's OK if you're dum https://dynomight.net/dumpy/

I have spent entirely too much time this week writing sorted list search algorithms (>2x speedup over binary search on triangle-finding with this one, though!). Time to go write some parsers instead.

Why do some people have a shit time with LLMs for programming while others love it?

What the latter group does is create tons of scaffolding to properly restrict LLMs, and acquire new skills to cope with all the unpredictable stuff they might do.

And because people are amazing at adapting the tools they’re given and totally underestimate the extent to which they do it, this is mostly invisible.

In this post, I argue that the amount of skill and scaffolding we grow is an incidental consequence of how badly AI coding tools are designed.

If we spent less time congratulating machines for the work people do and we instead supported people more directly, we could get much better outcomes.

https://ferd.ca/the-gap-through-which-we-praise-the-machine.html

just remembered to post about this slightly old news: a paper i coauthored with @cbaberle, Karl Crary, and Frank Pfenning was accepted to FSCD:

Substructural Parametricity

https://chrisamaphone.hyperkind.org/papers/fscd25-substructural-parametricity.pdf

the goal of this work is to get stronger "theorems for free"-style reasoning with a general account of substructural logics. e.g. there's only one (canonical) value of the ordered type ∀α. α ↠ α ↠ α • α . we simplify/generalize prior work by parameterizing our logical relation with a resource algebra

@krismicinski @jonmsterling I'm less interested in "keeping jobs local" or "keeping the US on top" and more interested in robustness/flexibility/opportunity. The anecdote about mask manufacturing locally during lockdown hit me hard. Entirely lacking certain expertises locally prevents quickly reacting to changing circumstances, and makes it harder for good ideas to be quickly tested and iterated on because the idea can't meet the persons who can build it.

@krismicinski @jonmsterling that part irritated me, because to me it comes off as "America used to be on top, now we're not, that's bad!" when to me, it's good that the rest of the world catch up to the West. To the extent that a global economy requires specialization, that means certain specialties moving out of the US. Other countries getting "smarter" is not a bad thing.

From the TYPES mailing list, Ichiro Hasuo posts about open positions at assistant and associate prof level (but I think not generally in line for tenure) at NII in Tokyo. He says: "I can certainly speak about my experience: Research is the top priority here. Little teaching (~ one course every two years for me); no undergrad students. You can still recruit and supervise grad students. Speaking Japanese is not needed at all (though it'll help your daily life). I enjoy scientific freedom. Yes, freedom." In case there are any refugees out there looking for at least temporary respite...

https://lists.seas.upenn.edu/pipermail/types-announce/2025/012009.html

@krismicinski @jonmsterling yeah, he's definitely got a political bias, and it comes through even in this video. but if you want to know why manufacturing in the US is difficult and costly, listening to people who've tried it seems relevant. unsurprisingly, those are disproportionately people who have ideological reasons to try to do so.

@jonmsterling I'm not sure what to think of this. My prior belief is that copyright & patents are mostly bad & that innovation happens despite IP not because of it. Prima facia this is a counterargument, but thinking coldheartedly it actually seems to support my prior: good designs get copied and are available at cheap prices; arguably the most innovative place in the world right now is one with lax IP protection. Still not fully happy with this conclusion.

@jonmsterling they also blame (~8m15s) China's disrespect for IP, suggesting that if you send CAD designs to China to manufacture, the factory may make cheap clones with the same tooling. They say a cofounder's product was cloned&undersold on Amazon this way.

@jonmsterling while I'm linking to youtube, here's a 1st-hand report from someone manufacturing a grill scrubber in the USA on what made it difficult: https://www.youtube.com/watch?v=3ZTGwcHQfLY

it's consonant with what you say - costs are too high - but also points to generational knowledge decay, expertise that is dying out along with older workers in eg the tool-and-die industry. China & India are centers of growing expertise by contrast. making stuff takes hands-on knowledge that is slow to travel, it seems.

@secretasianman @chrisamaphone @jonmsterling the video also discusses apples, which the US doesn't import much, but which are seasonal, so they must be kept from ripening in nitrogen-filled cold rooms for almost the whole year;

and table grapes, which can't be preserved nearly as long, so they alternate between different locations (California, Mexico, Chile) with different growing seasons and use varieties with extra-long harvest seasons.

@secretasianman @chrisamaphone @jonmsterling this video I saw recently goes into some detail about the logistics of fresh fruit availability in the US, including bananas & apples: https://www.youtube.com/watch?v=NmhDcZHg7ik (or on Nebula: https://nebula.tv/videos/wendover-the-logistics-of-yearround-fruit-supply)

it seems different kinds of fruit require different approaches. bananas grow year-round in the tropics, so the difficulty is shipping them. So: refrigerated containers on dedicated ships to avoid ripening until arrival.

@jonmsterling these seem like reasons manufacturing is good, not why it would be particularly good for it to be in the US as opposed to elsewhere.

again thinking pretty naively, to the degree manufacturing mostly happens in less prosperous places than the US, it seems like an argument _for_ off-shoring.

@jonmsterling

> re-shoring manufacturing would be a very good thing

naive Q: why would it be a good thing?

"Into the deep". The function
sin((1-v)·x) - sin(v·x + pi/3)
when v goes from 0 to 1.

#Mathart #algorithmicArt #creativeCoding #artwork #visualart #AbstractArt #NotAI

@dysfun hm, mayyybe shaved like 10s off the time (down to 1m50s in one run)? but that could be within the noise level, I see variance of 10-15s between runs I think.

@dysfun tried it, no observable difference. which make sense, I think? there are several types implementing the trait involved, and they're all generic, and the methods fairly small, and I've heard that generic things are basically always available for inlining.

(also there's no next() method, only seek() and position(), but I just applied #[always(inline)] to both).