huge wrench in my plans. python only supports building against openssl which is a much larger surface area than curl with gnutls
Discussion
Loading...
however i'm realizing this is may also possibly be exactly what i want. because while python's pathlib is easy the best vfs interface, its networking protocols are outrageously unfit
this spans the range from parsing json, to url quoting, to its intentionally broken "email" parsing, to its disastrous reliance on implicit context like the User-Agent string sent to pypi upon every request, which after 5 years of working on pip came as a revelation to me
a revelation in the most literal sense that the word admits. i have spent many hours and finally succeeded at deriving the simplest version of a type system that mypy and corporate greed stole away from me all those years ago. it ends up looking like pydantic on the surface but with significantly more care for crafting the transitions between states such that an unambiguous unidirectional data flow is achieved and described
this is significant because it is precisely such transitions which the pants v2 engine fails to provide language to express. instead, it has synchronous and asynchronous (coroutine) paths in the graph. it has no conception of resource allocation or RAII. it has no conception of failure, and therefore not of trying again
this is not because we had failed to consider whether such affordances would enable safer faster more robust build tasks. Retry (a wrapper over a path in the task graph which could be separately tracked) was one of the several incredibly compact prototypes i developed (along with the composable --query syntax) which were poorly received
it remains shockingly clear that the model of pure functions alone cannot hold. the embarrassing parallelism for scala compiles that win wang and i developed the first successful model for was very specifically enabled by codifying two separate forms of persistent state, locally and remotely, allocating a local jvm jit to run eugene burmako's rsc and allocating remote scoot workers from drew gassaway's squadron within a compile context that enabled them to temporally and spatially share resources for the so-called "ephemeral" scalac workers
~131 more replies (not shown)
it's very important that the task graph is able to represent more than pure functions because in fact literally every single i/o operation involves negotiating shared access to a limited resource. this is what makes it an i/o operation.
when i read that the zstd project categorically refuses to consider any benchmark outside of purely in-memory operations (they say this on the github in PRs, they're very open about it and it is used to justify extremely precise modifications to the code) was when i realized that project is not worth forking
resource pooling is not really an optimization technique so much as it is the bare minimum prerequisite to having any idea of what the system you've built is running up against. "explicit vs implicit" is far too shallow of a distinction to do justice to the difference in behavior of e.g. a rust program doing i/o with tokio's threaded async executor, vs a rust program that directs each thread to perform its own hot loop and simply sleeps until there's more data https://github.com/zip-rs/zip2/pull/236
coroutines themselves are not the problem. in fact it's well understood that ring buffers from the os are not just mario pipes but indeed serve the purpose of flow control with "backpressure". through coffeescript and atom-shell (electron before it became evil) i found node.js to have developed a wonderfully rich appreciation across the community for how to write in essentially continuation-passing style. it's a big relay race! we're doing it together! no wonder rust did the big rewrite from libuv barely 10 years ago!
but there is another presence in this world. not another tarnished that the light of grace would touch as prophesied by the elden ring cutscene, whose adventure begins at level 5, who uses a mirror to change their gender on demand
james baldwin wrote this 64 years ago https://www.newyorker.com/magazine/1962/11/17/letter-from-a-region-in-my-mind and it will haunt me long after the unix epoch transgresses 64 bits in seconds dulling our collective memory of the nixon presidency
Every effort made by the child’s elders to prepare him for a fate from which they cannot protect him causes him secretly, in terror, to begin to await, without knowing that he is doing so, his mysterious and inexorable punishment. He must be “good” not only in order to please his parents and not only to avoid being punished by them; behind their authority stands another, nameless and impersonal, infinitely harder to please, and bottomlessly cruel.
this sentential clause is written into the folds of my cranium— not burned or etched, but cryptographically signed into the ROM sector, the inner workings of it exposed by a clever reverse engineer
He does not know what the boundary is, and he can get no explanation of it, which is frightening enough, but the fear he hears in the voices of his elders is more frightening still. The fear that I heard in my father’s voice, for example, when he realized that I really believed I could do anything a white boy could do, and had every intention of proving it, was not at all like the fear I heard when one of us was ill or had fallen down the stairs or strayed too far from the house. It was another fear, a fear that the child, in challenging the white world’s assumptions, was putting himself in the path of destruction.
A child cannot, thank Heaven, know how vast and how merciless is the nature of power, with what unbelievable cruelty people treat each other. He reacts to the fear in his parents’ voices because his parents hold up the world for him and he has no protection without them.
the following takes the form of a thought experiment: perhaps of the form a nazi physicist might propose, in which he claims that the stochastic behavior einstein challenged until his death is responsible for whether a creature placed into his care lives or dies. schrödinger suddenly shows up with a flask of poison and starts demanding einstein take him seriously
have you ever looked up which poison he proposes for the act? it's a very specific poison. there are many poisons you could imagine might be able to kill a cat. but to erwin schrödinger the cat is largely immaterial to the entire enterprise
recall again that this is in response to einstein's scientific paper with two other jewish scientists https://en.wikipedia.org/wiki/Schr%C3%B6dinger's_cat
One can contrive even completely burlesque [farcical] cases. A cat is put in a steel chamber along with the following infernal device (which must be secured against direct interference by the cat): in a Geiger counter, there is a tiny amount of radioactive substance, so tiny that in the course of an hour one of the atoms will perhaps decay, but also, with equal probability, that none of them will; if it does happen, the counter tube will discharge and through a relay release a hammer that will shatter a small flask of hydrocyanic acid.
wikipedia has his original text in german inline as a citation, which unambiguously describes hydrocyanic acid, using the standard scientific terminology https://en.wikipedia.org/wiki/Hydrogen_cyanide#As_a_poison_and_chemical_weapon
Perhaps its most infamous use is Zyklon B (German: Cyclone B, with the B standing for Blausäure – prussic acid; also, to distinguish it from an earlier product later known as Zyklon A),[72] used in the Nazi German extermination camps of Majdanek and Auschwitz-Birkenau during World War II to kill Jews and other persecuted minorities en masse as part of their Final Solution genocide program.
none of the "interpretations" that are given scientific credence represent einstein's interpretation of the EPR thought experiment, which they devised and published together not as a philosophical quandary but a measurement of physical reality
when i attempted to request on the talk page where the work first received the derogatory term "paradox", i had my very minimal edits reverted with extreme prejudice https://en.wikipedia.org/wiki/Einstein%E2%80%93Podolsky%E2%80%93Rosen_paradox a day later the most obnoxiously authoritative replier followed up to say he believed it was schrödinger who said it. my addition of a template to request a source for this non-neutral point of view remains effectively censored. this is what einstein worked on for decades. it's not nothing
my thought experiment will make no claim to rival james baldwin nor albert einstein, nor their searing observation of what may be fancifully conflated to constitute the celestial mechanics of white supremacy. this name invokes the theory of epicycles, a historically controversial approach to orbital mechanics which baby physicists are taught as one of their first worked examples motivating their modern smug superiority over the past https://en.wikipedia.org/wiki/Deferent_and_epicycle#As_an_example_of_bad_science
galileo is considered a sympathetic figure in the history of epicycles. i was specifically told in school that he was "excommunicated" from the church, which i was informed constituted the harshest possible punishment. that appears to have been false, but what's more interesting is what incites that response. https://en.wikipedia.org/wiki/Galileo_affair
so the pope was quite literally a galileo patron and gave him explicit consent to publish on geocentrism "as long as he treated it as a hypothesis". given that telescopes in 1632 were limited to areas of clear sky with visible light and the robustness of ground glass (glass was still the weakest link for the hubble space telescope 400 years later amidst the cold war), the pope's admonishment is less harsh than modern funders of research into the dynamics of slop machines
but galileo had decided any resistance be it philosophical or even scientific critique was indicative of what he deemed an inferior intellect. he would go a lot further than needed with this. it was clearly really important to him:
Galileo complained that some of the philosophers who opposed his discoveries had refused even to look through a telescope:[11]
slopscope. teleslop
he would write a treatise regarding "the two chief world systems" https://en.wikipedia.org/wiki/Dialogue_Concerning_the_Two_Chief_World_Systems i mentioned the cold war earlier. bear this in mind. galileo would be wrong in many ways here, such as his impressive disregard for non-circular orbits. elliptic curves are more complex—and correct. what gets him into hot water was when he named a character after an ableist insult and used them to represent geocentrism
the term a priori is used here to trash aristotle. the only time i've ever seen that phrase italicized before in my entire life is the 1977 ziv and lempel compression paper which aggressively fails to cite shannon, claims to be independent of probabilistic assumptions, and is responsible for compression slop machines that are now the only type of compression anyone fucking uses
He adds "we possess a much better basis for reasoning about celestial things than Aristotle did...Now we, thanks to the telescope, have brought the heavens thirty or forty times closer to us than they were to Aristotle, so that we can discern many things in them that he could not see;"
thirty or forty times. this guy can't be bothered to read kepler's book kepler had personally sent to him which calculates the elliptical orbit of mars because he thinks seeing "30 or 40 times closer" is enough to start writing about the two competing world systems. elon musk behavior
in the immortal words of robert pike https://go.dev/talks/2012/splash.article "it's still bigger than we expected"
For the record, we measured the compilation of a large Google program written in Go to see how the source code fanout compared to the C++ analysis done earlier. We found it was about 40X, which is fifty times better than C++ (as well as being simpler and hence faster to process), but it's still bigger than we expected.
40X, which is (as we all know) fifty times better than C++
i find galileo's lack of care for elliptical orbits and deployment of what i believe to be correctly translated as an ableist epithet (wielded against the pope who specifically signed off on him publishing about geocentrism!) to represent a particularly detestable breed of physicist
While this story has been retold in popular accounts, there is no account by Galileo himself of such an experiment, and it is generally accepted by historians that it was at most a thought experiment which did not actually take place.[215]
Galileo was one of the first modern thinkers to clearly state that the laws of nature are mathematical.
here is why i believe the mathematician david hilbert to have been undergoing a literally fascist project and consequently how gödel's manipulation of his formal system describes why i believe formal systems can serve as resistance, when they can undermine hierarchical and even distributed power that seeks to lie with statistics—i.e. to perform eugenics
our orbit of galileo is complete. let us rewind the stack
the reason quantum gravity has been unsolved for a century and counting is because nazis and their apologists really hated that a jewish scientist can curve their spacetime. colonizers cannot envision the curvature of spacetime without thinking about how much more land they can own if they control it
this is why ziv and lempel (israeli mathematicians) refused to compare their work to existing universal coding schemes—because metrics of comparison are actually much more powerful than you would think. what's evil are benchmarks—a mockery of the statistical and mathematical and deeply empathetic analyses that link numbers to context. that most scientists and programmers directly rely upon numerical analyses that are often largely opaque to them is an intentional strategy to replace them. ziv and lempel sought to replace shannon entropy, and succeeded in creating compression slop. their novel contribution was enabling suboptimal behavior
but why did james baldwin haunt me here? https://circumstances.run/@hipsterelectron/116074111379910807
because javascript is not a programming language in the sense we usually imagine it to be
there is a reason that node.js first achieved an ecosystem of continuation passing in practice. there is also a reason why es6 syntax wears the skin of coffeescript while webassembly removed the todo for the goto operator in december 2024. there is a reason es6 imports are broken for file:// imports. there is a reason why they are only supported at an entirely new scoping level https://github.com/jashkenas/coffeescript/pull/5475
there is a reason why the async framework in javascript has exactly one executor and it's not you. there's a reason why rust's async model is completely broken for any task in which you care about hot loops and task switching. there's a reason astral refuses to use my rust zip implementation and why fasterthanlime made a 30 minute video a year and a half ago which would preface astral's campaign to take credit for my work on pip while inculcating the factually incorrect claim that async is more efficient in any way shape or form
the reason why james baldwin spoke to me in this moment where i found context to be the missing piece that had been kept from me previously is the part immediately after what i shared above
I defended myself, as I imagined, against the fear my father made me feel by remembering that he was very old-fashioned. Also, I prided myself on the fact that I already knew how to outwit him.
rust names its projects 'nushell', the way 'nu disco' is so named to differentiate itself from disco. one wonders if the egg can crack from the shell. one wonders how to describe a linguistic creole in terms that would be acceptable to imperialists. computers refuse to spell my last name right because it's from one of the languages the english didn't succeed in erasing all traces of yet. the most exciting work i've seen on the shell front is from openrc which deftly wields the twin forces of portable code: a shell, and a C compiler.
"stop parsing strings and start solving problems" is a deeply populist and farcically anti-intellectual slogan. all interesting problems are parsing strings and translating between dialects. the shell environment persists because process executions are a lingua franca, the fabled N-to-M FFI. if you're parsing strings that means you're talking to a human. if you're parsing output that wasn't meant to be stable, is that because the tool doesn't provide a stable CLI, or because you're trying to subvert or replace the tool in an adversarial way?
i posted one email to the findutils mailing list with two typo fixes and they asked for my preferred name and email address to credit me. nushell describes implicit returns as "similar to powershell or rust" https://www.nushell.sh/book/thinking_in_nu.html#implicit-return
it's lisp. you mean lisp. maybe you can remember it better with this mnemonic: LInguiStic imPerialist
To defend oneself against a fear is simply to insure that one will, one day, be conquered by it; fears must be faced. As for one’s wits, it is just not true that one can live by them—not, that is, if one wishes really to live.
i believed myself to be striving for success in sticking with projects and programs and people who are uninterested in the science of measurement, who wield benchmarks like bulldozers over stolen land, who withold compensation prior to a satisfactory level of debasement, who maintain that low-level grinding against AI enemies is the only way to level up, to become like them, to arbitrage their nightmare to a greater fool. i believed to dream was to initiate the process of regret.