This is possibly one of the more cursed single sentences I've ever seen in a job posting

[aud@fire.asta.lgbt]

@[email protected] holy shit. That's what Unity does?! Oh my god.

[xgranade@wandering.shop]

@aud Yep! IL2CPP! It's not not even the most ridiculous, either. Maybe they'd have gone with LLVM IR or Cranelift IR starting from scratch now, but CIL is a pretty good IR — it wasn't designed for interfacing between different modular stages of a compiler toolchain like LLVM IR and Cranelift IR, but you can absolutely hit it with a hammer until it does the job.

Add that a lot of console SDKs work well with C++, and it's a hell of a hack, but it solves a real problem.

[xgranade@wandering.shop]

@aud (The distinction here being between IRs that were designed for interpretation and/or JITing, like JVM bytecode and CIL, and those IRs that are meant to be amenable to compiler passes, like LLVM IR and Cranelift IR. You tend to see stack-based for the former and the latter tends to use SSA on an infinite set of registers.)

[aud@fire.asta.lgbt]

@[email protected] I wonder if there is a single "problem" out there trying to use an LLM that isn't just a "this is already a solved problem, we just find the whole 'paying livable wages thing' annoying" ploy.

[xgranade@wandering.shop]

@aud Sorry for infodumping again. Worked a bunch on IRs for quantum stuff, which inevitably (because error correction requires Turing-complete execution) requires at least as much IR infrastructure as for classical execution.

[odoruhako@mastodon.social]

@aud Sounds like a fun survival horror game.
But as a job, just no!

[aud@fire.asta.lgbt]

@[email protected] oh no, this is really cool! Please never apologize for discussing things. There's a lot of things you know that others don't, and the chance to learn or hear about new stuff isn't an opportunity you should feel bad for giving people!

(especially me, I know rather little about this particular domain)

[peter_mcmahan@mas.to]

@schrotthaufen @blogdiva @aud cispiler... so `cat`?

[aud@fire.asta.lgbt]

@[email protected] "it transpiles the COBOL into the JAVA with the LLM or it gets the hose again!"

[xgranade@wandering.shop]

@aud There's that, I agree, but there's also the problem that people who don't know the problem space also don't know what solutions already exist, nor how to evaluate the fitness of a proposed solution.

That's fine as long as you stay curious and all, but that's not the VC techbro way. All of a sudden, LLMs allow them to *appear* to be competent on a wild array of different things that are *already solved*.

[xgranade@wandering.shop]

@aud Like, the difference between "this is a trivial problem I'd assign an undergrad," "this is a tricky problem, but solved; applying the solution takes a lot of work," "this is unsolved, but you can kind of hack around it," and "this is unsolvable, ever, as a matter of principle" is often pretty subtle and takes some training?

[aud@fire.asta.lgbt]

@[email protected] Yeah, I suspect that's the motivation for a lot of them.

I fully admit I sent in two applications today where the subtext of my language was "This application is conditional based on the idea that you aren't just trying to shoehorn the spiky club that is an LLM where it has no business being and are actually interested in solving the problem you claim to be interested in".

[xgranade@wandering.shop]

@aud

- Will this program ever crash? Impossible.
- Will this program ever crash due to a memory safety violation? Difficult, but solvable.
- Does this program include any infinite loops? Impossible.
- Does this program include something from this list of common kinds of infinite loops? Trivial.
- What's the smallest convex shape that includes these points? Solved, but slightly tricky.
- How can this UPS truck take the least number of left-hand turns? I hope you have a galaxy-sized computer!

[xgranade@wandering.shop]

@aud

- What's the solution to this polynomial equation? Trivial.
- Does this polynomial have an integer-valued solution? *cries in agony*
- Does this list of two-variable constraints have a solution? Trivial.
- Does this list of three-variable constraints have a solution? *cries in agony*
- Does this list of two-variable constraints have an approximate solution? *cries in agony*
- Is there a solution to this conda env file? *cries in agony*

[xgranade@wandering.shop]

@aud Like it's just so fucking easy to take a problem in P and accidentally turn it into the halting problem!

Fucking grep did that! By adding a backreference feature that meant that a regex was no longer actually decidable in finite memory, but was undecidable in general!

[xgranade@wandering.shop]

@aud "Is this string a valid e-mail address?" Depends on what you mean by "string," "valid," and "e-mail address," but it's probably either trivial or *cries in agony*.

[xgranade@wandering.shop]

@aud (I'll stop now, I promise.)

[aud@fire.asta.lgbt]

@[email protected]

Does this program include any infinite loops? Impossible

is this one of these "you can't create an algorithm to solve this for any and all loops but could do it for certain cases" type problems?

[aud@fire.asta.lgbt]

@[email protected]

Is there a solution to this conda env file? cries in agony

lmao

[aud@fire.asta.lgbt]

@[email protected] This reminds me of how apparently quite a few job postings have a required LinkedIn profile field and so I've had to cheekily write in https://linkedin.com because apparently they do some string parsing so that lmao://nope.no doesn't work but.