zwarich
zwarich@hachyderm.io
<p>Programming language & compiler enthusiast, computer architect. Creator of Rosetta 2.</p>
Posts
-
View post
efficient dense low-precision linear algebra is The Great Filter
-
View post
You know you’re getting old when you remember remembering something
-
View post
Algol-68 unions: https://en.wikipedia.org/wiki/Denmark–Norway
-
View post
After watching at the &amp;Truth intro video (https://www.youtube.com/watch?v=OU29x5TZ0w0), it seems clear that they are attempting to &quot;verify&quot; so much SW that it is no longer possible for humans to even read the function-level separation logic specs for the entire C ecosystem, never mind write them. The only logical conclusion is then that they are producing slop specs and then proving these slop specs using slop proofs. I really have nothing against slop proofs (and thin...
-
View post
this extra hour of sleep is great, we should add an hour every day
-
View post
Russian bot network are down, so we're free to criticize Perl 6 again
-
View post
too bad Nathan Fielder didn't do the C++ documentary
-
View post
After a long discussion on the Lean Zulip, Mario Carneiro has shown that the Calculus of Inductive Constructions (in its full version w/ large elimination of Acc) + Excluded Middle proves that ZF is consistent: https://arxiv.org/abs/2609.23143 It is fairly simple to show that Aczel’s type of sets as well-founded trees is a model of ZF without Replacement, and even a bit further, that the iterative hierarchy V_alpha exists for all ordinals alpha, but Replacement seems very analogous to Unique Cho...
-
View post
After playing around with various attempts to construct models of ZF (or IZF) in the Calculus of Inductive Constructions, I can say that CiC (even with LEM) is severely deficient compared to ZF/IZF in terms of what can be done without assuming some form of Choice. I think that Lean&#39;s banishment of these problems by just assuming Choice throughout the ecosystem might be no small part of its appeal to mathematicians.
-
View post
I don&#39;t even eat dumplings anymore, I just buy into a dumpling REIT
-
View post
@joe@f.duriansoftware.com Zeroing is adding a new "row" syntax, motivated by token efficiency: type Point x 132 y 132 fn sum i32 point Point ret + point.x point.y pub fn main Void world World ! let point Point . x 40 y 2 let total sum point if == total 42 check world.out.write "ok\n"
-
View post
I’ve seen a number of people speak about a representation of datatypes in ML where a datatype is given by a module with an abstract type and its constructors. Obviously, this is missing some additional information to define pattern matching, e.g. a recursor. Has anyone actually worked out all of the details?
- View post
-
View post
The latest Lean soundness hole (https://github.com/leanprover/lean4/issues/14576), which also affects alternative implementations of the kernel, is proof that God only ever intended for us to use W-types, and certainly not nested inductives.
-
View post
we need to consider deceleration
-
View post
@joe@f.duriansoftware.com @slava@mathstodon.xyz I know we talked about this as part of a shitpost thread, but can either of you think of a good way to combine traits (i.e. type classes) and bytewise equality / hashing while still preserving as much parametricity as possible? The natural thing is to have some marker trait with C++ style specialization, but that destroys a lot of the properties that you would want. @joe@f.duriansoftware.com mentioned (jokingly but maybe not) to just make it requi...
-
View post
@dotstdy@mastodon.social @ianh@mastodon.social Has any recent language adopted uncurried single arg style where product types are used for multiple args rather than implicit currying? I know that Swift originally started in this direction, but I forget the actual reason why they backed away from this. Can you refresh my memory, @joe@f.duriansoftware.com ?
-
View post
@joe@f.duriansoftware.com Why did Swift never get block expressions? Was nice trailing lambda syntax worth more?
-
View post
@joe@f.duriansoftware.com That's for humans. They have separate install instructions for agents using what appears to be another npm extension for agent "skills": $ npx skills add vercel-labs/zero
-
View post
I'm always curious about peoples' pure C compiler codebases, so I took a look at Vercel's new Zero language (https://zerolang.ai). You can tell that someone is just fed up with the bullshit of using unions in C when you see this: struct Expr { ExprKind kind; char *text; char *resolved_type; bool moves_ownership; bool mutable_borrow; bool bool_value; bool array_repeat; Expr *left; Expr *right; ExprVec args; TypeArgVec type_args; TypeArgVec checked_type_args;...
-
View post
@joe@f.duriansoftware.com Yeah, it seems pretty likely the C codebase is vibe-coded, although it’s interesting that it doesn’t have any comments. I wonder if that’s a deliberate context management strat?
-
View post
@regehr@mastodon.social My first reaction when seeing the Claude C Compiler was “John Regehr is going to be able to milk this so hard”.
-
View post
I treat garbage collected languages like Undertale and always go for a pacifist run.
-
View post
@joe It occurred to me while walking today that if all of the arguments for stackless coroutines are valid, then Rust should actually switch to CPS for compiling all (non-recursive?) functions.
-
View post
@joe@f.duriansoftware.com What do you think is the maximal subset of effect handlers that can be supported well in a C-like language (with a C-like compilation model)?
-
View post
GitHub is the only web app that got less reliable after it stopped being (as much of) a Rails monolith.
-
View post
@shac this was basically me at 4 AM debugging AArch64 iOS during UI bringup, except I'd just go home to sleep and let Yosen figure it out in the morning
-
View post
You either die a hero or live long enough to break subject reduction.
-
View post
I just saw an ad that began with “inflight Wi-Fi has come a long way since I was a kid”. I am officially unc, also chopped.
-
View post
@lynnesbian This but with gallium arsenide semiconductors instead of silicon CMOS.