English
Felipe O. Carvalho
27.5K posts
Felipe O. Carvalho
@_Felipe
@ApacheArrow / Databases / Compilers. (past @SDFLabs, VoDa, @Spotify). Rust/C++/TLA⁺🇧🇷 → 🇸🇪 → 🌎 https://t.co/vxbdByfADI
เข้าร่วม Haziran 2008
2.4K กำลังติดตาม4.9K ผู้ติดตาม
TIL: modern CPUs can can predict (cache) tens of thousands of branches. This makes invalid many benchmarks with small amounts of input data.
Daniel Lemire@lemire
Daniel Lemire, "How many branches can your CPU predict?," in Daniel Lemire's blog, March 18, 2026, lemire.me/blog/2026/03/1….
English
I can't even begin to describe how much that's not what I want.
English
@theodorvaryag @joseph_h_garvin Separate compilation is a fundamental language design dimension. It affects many aspects of a language and the compiler, not just build times. e.g. ability to load code dynamically.
English
@joseph_h_garvin @_Felipe Honestly I don't know why I keep trying to give alpha away to people who lend me zero charity, extend no inferential distance.
Forget I said anything. Nothing to learn here. Continue as you were!
English
4 words in and I knew this was a Haskeller Rust takedown. It’s a good thing that Rust doesn’t have HKTs, GADTs, typeclasses…
Jules@analytichegel
go watch this rust takedown, it's pretty good. youtube.com/watch?v=1iPWt1…
English
@kenwheeler @RandallKanna Top skill: handling incidents while no one who pushed code to main during last sprint has no idea what they have done.
English
@RandallKanna i’ll give you a real answer here. all the orchestration and tweaking is pure masturbation. the value moving forward is identifying places you can implement this deterministically in a clever way to provide value. it’s that simple. but at the same time not simple at all.
English
Honestly what is there to compete on at this point for a dev job? Like code reviews?
English
@joseph_h_garvin typeclasses can be virtual so code that uses them can be compiled separately. Rust traits require specialization unless dyn-compatible and code opts in to it. Less flexible, better inlining and optimization. typeclasses don’t always preventing inlining but they don’t require it.
English
@_Felipe I mean, it does have type classes though, they're just called traits. Some of these I can tell are bad just from the screenshot, e.g. Hoogle mostly exists to make it possible to search for custom binary operators a problem Rust just doesn't have.
English
@ForrestTheWoods You can use a mouse on tmux:
set mouse on
Normal people don’t use TUIs. That’s why TUIs don’t work for normal people.
English
Unpopular Opinion: tmux controls are absolute dog 💩 🗑️. TUI controls don’t have to be garbage. You can just make normal controls that work for normal people. And let users use a mouse cursor. It’s totally fine and good actually.
vim people I swear. The worst.
English
@macrocephalopod @investingidiocy It works if they are similar too
English
@investingidiocy I am begging you to realise that the whole point of this meme is that the left and right sides say the same thing 😭
English
@ID_AA_Carmack I have friends that work on audio streaming/decoding at Spotify and Bluetooth frustrates them like nothing else.
English
When you stream Spotify to Bluetooth speakers or headphones, the audio comes over the network lossily compressed with Vorbis or AAC codecs, is then decoded on your device to 48 Khz raw samples, then the Bluetooth stack lossily re-compresses it with SBC or AAC codecs before sending it over the airwaves to the speakers.
I don’t have “golden ears” to pick apart audio quality like I can with, say, missing gamma correction on texture filtering, but that still hurts my system optimization soul. It is likely over-optimization, but It would be cleaner if there were a way to send bluetooth-ready, compressed audio directly.
English
Unpleasant truth about programming: solutions that wind up winning aren't the cleanest, most elegant algorithms and design but rather the ones with (morally speaking) the most giant, complete and horrible set of switch statements covering all the weird and interlocking cases.
English
@StatisticsFTW @joseph_h_garvin LLMs would have to deal with combinatorial explosion of model which usually requires creative leaps and a feel for where failure modes might hide in an algorithm. Then after that, successfully refine the spec into an implementation.
English
Even Rust is not enough to rescue agents from building software with flaky parallelism. It can prevent thread races from causing memory unsafety but that's it. Unless there's some magical prompting trick that makes LLMs better at reasoning about multithreaded code?
English
@rsms @v1gnesh @_Felipe +1 that queueing theory may be the best reference; good motivation in an old (2015) talk, "Everything Will Flow" by Zach Tellman: youtube.com/watch?v=1bNOO3…
Some good refs. in "Queueing Theory, In Practice" by Eben Freeman, speakerdeck.com/emfree/queuein….
YouTube
English
Is there no good canonical article on "Backpressure" applied to Computer Science? The wikipedia one talks about fluids and pipes. It's related but bad as a citation.
English
@mgill25 Any kind of resource management where one component needs a resource and the more it gets the resource the more it needs. e.g. the more memory you give to a task scheduler, the more tasks it will spawn that will also need memory.
English
@AgileJebrim "requiring a subset of code be used that has hard bounds on execution times" Right, you have to restrict the construction of the program to be able to verify it halts.
English
This is utter nonsense of course. The entire basis of the field of real-time computing is predicated on guarantees that computers will complete within a given deadline. We just bypass Turing’s proof by requiring a subset of code be used that has hard bounds on execution times.
vixhaℓ@TheVixhal
You cannot write a computer program that looks at any other program and correctly determines whether it will eventually finish or run forever. This isn't a hardware limitation. Alan Turing proved in 1936 that no such program can logically exist, by showing any attempt creates a paradox, which is essentially a formal version of “this sentence is false.” What's unsettling is the proof technique: he didn't find a hard case. He proved the question destroys itself when you try to answer it.
English
@dillon_mulroy I’ve kept copilot on nvim since forever. It’s not very smart but does the obvious stuff well.
English
i think i’m back to wanting a really good tab model - any progress here outside of cursor (i don’t have access to supermaven) and for nvim?
English
@weswinder And strangers online disagreeing are all just “haters”.
English
the problem with marc, garry, elon, etc
is everybody just wants their money
so nobody in their life is willing to tell them when they are wrong because they are afraid of not getting money
and they end up with a bunch of sycophantic yes men in their circles (powered by ai)
English
@LcavalcanteR @marcogomes O iCloud é um espelho do telefone. Se vc não quiser no iCloud, move do telefone pra um PC.
Português
@marcogomes Puts. Eu não fui claro, perdão.
Eu não quero perder as filmagens, preciso dos vídeos em um HD, de forma local.
Eu vacilei em ter salvo no telefone, tenho quase 100GB que não consigo pegar
Português
esses dias comentei que o iCloud synca quando quer, e algum corno veio dizer que eu estaria mentindo porque >pra ele< funciona.
tô esperando essas 110 fotos subirem faz aproximadamente 40min (que subisse pelo menos 1, mudando para 109)
Português
@wagslane They are usually smart. I avoid *saying* the word and favor *unrelated*.
English
still trying to decide if I should avoid or be drawn to people who say "orthogonal"
English