Sign in

Alice ✨

@welltypedwit.ch
3K followers 343 following 6.6K posts

There’s no sense crying over every mistake. You just keep on trying ’til you run out of cake - GLaDOS welltypedwit.ch

PostsRepliesMedia
Reposted by Alice ✨
PolyWolf @wolf.girl.technology · 29/09/2026
New blog post: "Pining for Arc Downcasting in Rust" wolfgirl.dev/blog/2026-09-29-pining… #programming #rust
6243
Alice ✨ @welltypedwit.ch · 27/09/2026
funny how if you pick the right benchmark, ghc can beat rustc by a factor of 10
1120
Alice ✨ @welltypedwit.ch · 27/09/2026
i love tailcc
$ time vega build -O3
Executed in  365.52 millis
...
$ time vega build -O0
Executed in    7.32 secs
080
Alice ✨ @welltypedwit.ch · 26/09/2026
oh that stupid incorrect haskell video is made by the guy that made that stupid incorrect rust video. that makes sense.
6360
Alice ✨ @welltypedwit.ch · 25/09/2026
> In addition, premature promotion will increase the incidence of nepotism, as ‘tenured’ garbage objects in the old generation preserve their offspring in the young generation, artificially inflating the survival rate as those dead children will also be promoted. we have... interesting terminology
1400
Alice ✨ @welltypedwit.ch · 24/09/2026
for all their flaws, sometimes i really wish i could have either an ocaml-style interface file or haskell-style export list in rust. it's so nice to be able to specify the interface cleanly in one place before worrying about the actual implementation
3280
Reposted by Alice ✨
waffle .-. @waffle.pet · 23/09/2026
I'm giving a talk titled "breaking changes are a social construct" tomorrow at the Amsterdam Rust meetup. You should come if you can! www.meetup.com/rust-amsterd...
meetup.com
Rust Meetup @ BlockTech, Thu, Sep 24, 2026, 6:30 PM | Meetup
Join us for an evening of cutting-edge Rust talks, real-world use cases and great conversations hosted at BlockTech. This meetup is open to anybody who has an interest in t
2354
Alice ✨ @welltypedwit.ch · 21/09/2026
so, are there any "philosophy of ai" papers (or papers about fundamental limits of ai/llms in general) that aren't... pseudoscience? I think all the ones I've read went something like this:
4192
Alice ✨ @welltypedwit.ch · 20/09/2026
it's funny how ocaml people always go on about how much ocaml focuses on backwards compatibility and how they can't fix any of its major problems because they need to keep old code compiling but also like 10% of all functions in a project of mine broke because `effect` became a reserved keyword
3404
Reposted by Alice ✨
Kiran @kirancodes.me · 18/09/2026
I like how the author has pre-emptively piss-coloured the paper in respect for its contents
BendRT: a parallel runtime for CPUs and GPUs
5381
Alice ✨ @welltypedwit.ch · 18/09/2026
if ghc 9 is so good why isn't there a- discourse.haskell.org/t/ghc-10-0-1...
discourse.haskell.org
GHC 10.0.1-alpha1 is now available
The GHC developers are very pleased to announce the availability of the first alpha prerelease of GHC 10.0.1. Binary distributions, source distributions, and documentation are available at downloads...
2181
Alice ✨ @welltypedwit.ch · 18/09/2026
streaming platforms should have non-temporal loads
180
Alice ✨ @welltypedwit.ch · 17/09/2026
i made some major changes to polaris for the first time in a while and wow I did not miss the user experience of writing ocaml lol
1170
Alice ✨ @welltypedwit.ch · 12/09/2026
you know, i thought the reputation of cousot papers was at least a little exaggerated but no, they actually look like that lol (this is page 3)
a page full of symbols and operators and... biblically accurate expressions spanning several lines casually inlined into regular sentences as if they were obviously understandable
8609
Alice ✨ @welltypedwit.ch · 12/09/2026
omg do people call the concretization function in galois connections "γ" because γ is the third letter of the greek alphabet so it's kind of the equivalent of "c" (as in *c*oncretization)?
2120
Alice ✨ @welltypedwit.ch · 12/09/2026
you know it's not a great look that every single atproto tool or app besides bsky itself is fully vibecoded, complete with ai generated text
5320
Alice ✨ @welltypedwit.ch · 11/09/2026
pro tip: if you're going to faint, try not to hold anything made of glass
3221
Reposted by Alice ✨
Kiran @kirancodes.me · 09/09/2026
C++ is a great eldritch horror; the only thing it's missing is a halfway decent programming language
2263
Alice ✨ @welltypedwit.ch · 09/09/2026
this one is going to stay relevant for a while huh
2310
Alice ✨ @welltypedwit.ch · 06/09/2026
omfg llvm is looping again when i use tailcc
160
Alice ✨ @welltypedwit.ch · 06/09/2026
as much as i love vega, polaris has by far been the most impactful thing i've ever made (for myself). it's literally the only language i know that makes scripts actually pleasant to write.
1100
Reposted by Alice ✨
gabby @fullmoon.id · 04/09/2026
Look at what Haskell programmers have been demanding your Respect for all this time
(<<$>>) :: Functors '[f1, f2] => (a -> b) -> f2 (f1 a) -> f2 (f1 b) infixl 4 
(<<<$>>>) :: Functors '[f1, f2, f3] => (a -> b) -> f3 (f2 (f1 a)) -> f3 (f2 (f1 b)) infixl 4 
(<<<<$>>>>) :: Functors '[f1, f2, f3, f4] => (a -> b) -> f4 (f3 (f2 (f1 a))) -> f4 (f3 (f2 (f1 b))) infixl 4 
(<<<<<$>>>>>) :: Functors '[f1, f2, f3, f4, f5] => (a -> b) -> f5 (f4 (f3 (f2 (f1 a)))) -> f5 (f4 (f3 (f2 (f1 b)))) infixl 4 
(<<*>>) :: Applicatives '[f1, f2] => f2 (f1 (a -> b)) -> f2 (f1 a) -> f2 (f1 b) infixl 4 
(<<<*>>>) :: Applicatives '[f1, f2, f3] => f3 (f2 (f1 (a -> b))) -> f3 (f2 (f1 a)) -> f3 (f2 (f1 b)) infixl 4 
(<<<<*>>>>) :: Applicatives '[f1, f2, f3, f4] => f4 (f3 (f2 (f1 (a -> b)))) -> f4 (f3 (f2 (f1 a))) -> f4 (f3 (f2 (f1 b))) infixl 4 
(<<<<<*>>>>>) :: Applicatives '[f1, f2, f3, f4, f5] => f5 (f4 (f3 (f2 (f1 (a -> b))))) -> f5 (f4 (f3 (f2 (f1 a)))) -> f5 (f4 (f3 (f2 (f1 b)))) infixl 4
86613
Reposted by Alice ✨
Jenna JJ Q. @tomatofacts3.bsky.social · 03/09/2026
ah the classic. "how the FUCK does rustc -- oh, it crashes"
0181
Reposted by Alice ✨
Alice ✨ @welltypedwit.ch · 03/09/2026
I'm starting to think I should rethink the way I was planning to do niche-filling optimizations in vega. I was initially going to do essentially what rust does: have a normal layout with byte-based offsets/alignments and include extra information about available niches
261
Reposted by Alice ✨
Alona @alona.page · 02/09/2026
dom mindset: every path to that node must go through me first switch mindset: i hope my case labels are consecutive so i can be transformed into a jump table sub mindset: i don’t care if my operands are signed or unsigned
0377
Alice ✨ @welltypedwit.ch · 03/09/2026
I'm starting to think I should rethink the way I was planning to do niche-filling optimizations in vega. I was initially going to do essentially what rust does: have a normal layout with byte-based offsets/alignments and include extra information about available niches
261
Reposted by Alice ✨
Max @bernsteinbear.com · 01/09/2026
New paper with the ZJIT team! Local variables in Ruby are ... complicated. bernsteinbear.com/assets/img/s...
bernsteinbear.com
1101
Alice ✨ @welltypedwit.ch · 01/09/2026
i don't actually believe this, but i really wouldn't be surprised if the data center water usage thing had been deliberately spread by openai or something like, literally replace "water usage" with "carbon emissions" and you have a great argument lol
3142
Alice ✨ @welltypedwit.ch · 29/08/2026
always fun when finding the source of a bug takes several hours and then fixing it takes 40 seconds but only because you spent 35 of them wondering what on earth is wrong with you
0573
Alice ✨ @welltypedwit.ch · 29/08/2026
i'm still really happy with vega's (non-contiguous) layout system but wow the generated llvm is horrible to read
190
Alice ✨ @welltypedwit.ch · 29/08/2026
uhh is this bad?
Fatal glibc error: malloc.c:2199 (sysmalloc): assertion failed: (old_top == initial_top (av) && old_size == 0) || ((unsigned long) (old_size) >= MINSIZE && prev_inuse (old_top) && ((unsigned long) old_end & (pagesize - 1)) == 0)
280
Alice ✨ @welltypedwit.ch · 27/08/2026
can llvm stop overriding its caller's stack frame please
1100
Alice ✨ @welltypedwit.ch · 26/08/2026
i love llvm
docs for a function "removePredecessor":

Note that this function does not actually remove the predecessor.
0400
Alice ✨ @welltypedwit.ch · 26/08/2026
so we're at the stage where "you should understand the code you're committing" is controversial now huh
4836
Alice ✨ @welltypedwit.ch · 26/08/2026
is there any real implementation of full linear (not affine) types that isn't haskell btw? (the haskell implementations has a few... rough edges... that aren't really inherent to linear types so i'm wondering if anyone has made them actually nice to use yet)
5140
Reposted by Alice ✨
amos @fasterthanli.me · 25/08/2026
Wow linear types suck if you contemplate the consequence for twelve seconds actually
8302
Alice ✨ @welltypedwit.ch · 24/08/2026
interesting feeling to see some code where i know because of the author that it's probably been vibecoded but it also does something so obviously stupid that i'm not sure if an llm would have produced something like that
4280
Alice ✨ @welltypedwit.ch · 22/08/2026
i kind of wish rust cared about modularity, like, at all
3150
Alice ✨ @welltypedwit.ch · 20/08/2026
it's amazing how even cabal is a better build tool for C/C++ code than cmake
5271
Alice ✨ @welltypedwit.ch · 20/08/2026
sometimes i think haskell and ocaml have bad tooling. and then I use literally anything else and am reminded that no, most languages are worse off actually lol
2381
Alice ✨ @welltypedwit.ch · 20/08/2026
man, bitsets are so cool, especially in combination with pointer tagging. like, wdym you can do unions, intersections and everything and as long as you have less than 64(!) elements, this is cheaper than an array access
3511
Alice ✨ @welltypedwit.ch · 19/08/2026
you would think there would be a point where i wouldn't have to look up how llvm's getelementptr instruction works on structs
1110
Alice ✨ @welltypedwit.ch · 19/08/2026
i know how to write c++
struct StackFrameAssignments {
    assignments :: DenseMap<Value *, unsigned>
};
0110
Reposted by Alice ✨
Haskell programming language @haskell.org · 19/08/2026
Does *your* runtime system embed the Eye of Sauron? Yeah, I didn't think so. 💅 #Haskell
Description of selector thunks in the GHC runtime system, and suddenly the sentence "the garbage collector sees all, knows all."
3606
Reposted by Alice ✨
PolyWolf @wolf.girl.technology · 19/08/2026
New blog post: "Given A Link To This Post, Can You Trust It?" wolfgirl.dev/blog/2026-08-19-given-… #cybersec
043
Alice ✨ @welltypedwit.ch · 19/08/2026
wait bun's ""11 day"" ai rust rewrite is still not out?
2220
Alice ✨ @welltypedwit.ch · 19/08/2026
F. the first one is literally a joke with the punchline removed. the "code snippet" thing is directly plagiarized from another post but with the code removed so that it doesn't make sense anymore. also, elixir? chiptunes?? where did that come from
Have you ever thought static type systems spent too much time preventing runtime crashes? Have you ever wished a missing type constraint could invoke undefined behavior? Then you should try writing an obscure parser combinator in Haskell while listening to chiptunes!Type systems: Debating whether to rewrite it in Elixir or just add 63 extra sections to the proof.Code snippet: test:Main:blah(f_0 : (FunctionPointer * Boxed), x_1 : ALT) — can anyone guess what function this implements? ^^Vibe: Smart enough to know better, dumb enough to do it anyway.If you want, tell me:What programming language or type system are we debugging?Should we add more abstract algebra or just give up and use a segfault?
120
Alice ✨ @welltypedwit.ch · 16/08/2026
today in "only in c/c++": iterating over blocks in reverse order removes most of the instructions from each block
190
Alice ✨ @welltypedwit.ch · 14/08/2026
isn't that a cute hack
{- | Even though we statically link with our plugin, stack does not
re-link the project when the plugin archive changes.

In order to force it to do so anyway, we use this hack:
We declare a template haskell dependency from an otherwise empty module
(Vega.ShadowStackDependency).
This way, whenever the archive changes, stack will recompile only that
module, which doesn't actually do anything but still makes it
re-link with the new version of the plugin.
-}
addShadowStackDependency :: TH.DecsQ
addShadowStackDependency = do
    TH.addDependentFile "llvm-shadow-stack/libVegaShadowStackPlugin.a"
    pure []
1171
Reposted by Alice ✨
Max @bernsteinbear.com · 13/08/2026
More SSA fun! bernsteinbear.com/blog/more-pa...
bernsteinbear.com
Another partial SSI trick with canonicalize
After reading Chris Fallin’s aegraph post, new ZJIT contributor dak2 landed a block-local version of the canonicalize function in #16828.
0112