Hacker News: mkehrt

New comment by mkehrt in "Dust: Pretraining Transformers Without Backpropagation"

mkehrt — Tue, 06 Oct 2026 17:28:38 +0000

I’m not sure that’s true. We have literal billions of years of training.

New comment by mkehrt in "Type Safe Generic Data Structures in C"

mkehrt — Mon, 05 Oct 2026 17:42:11 +0000

I absolutely cannot remember the details, but we did typesafe, macro based generic data structures in C in an undergrad class I took ~20 years ago (CMU Operating Systems). The idea has definitely been around, though I don't know if the implantation was the same.

New comment by mkehrt in "Claude partial outage"

mkehrt — Tue, 29 Sep 2026 16:26:15 +0000

Take it up with Isaac Asimov, I guess.

New comment by mkehrt in "Book review: Is parallel programming hard, and, if so, what can you do about it?"

mkehrt — Fri, 25 Sep 2026 00:08:45 +0000

Well, we were using it to talk about things like cache invalidation and lax memory models rather than properties of algorithms.

New comment by mkehrt in "Book review: Is parallel programming hard, and, if so, what can you do about it?"

mkehrt — Thu, 24 Sep 2026 22:08:09 +0000

As other comments have pointed out, this is just not true in general usage.

When I was a grad student studying this stuff (~20 years ago), we used "parallelism" to mean running on different cores at the same time and "concurrency" to mean preemptive multithreading on a single processor.

New comment by mkehrt in "Anecdotally, programmers dislike "reduce""

mkehrt — Thu, 17 Sep 2026 04:10:32 +0000

> You could literally "reduce" and array of integers to a cancellation token, or a state object, or anything else

This is what reduce does, though? It reduces a list to a single thing. It seems like you're thinking of filter.

New comment by mkehrt in "Small programming tricks"

mkehrt — Thu, 17 Sep 2026 03:48:56 +0000

You can disable this misbegotten feature with

stty -ixon

New comment by mkehrt in "Small programming tricks"

mkehrt — Thu, 17 Sep 2026 03:47:48 +0000

This is a readline feature. You can enable it in bash (and other shells?) with:

bind '"\e[A": history-search-backward'

bind '"\e[B": history-search-forward'

New comment by mkehrt in "Small programming tricks"

mkehrt — Thu, 17 Sep 2026 03:43:55 +0000

Interesting; I assumed the write to stdout would just hang for the writing program.

New comment by mkehrt in "Rust is tier-1 language at Microsoft"

mkehrt — Fri, 11 Sep 2026 00:25:16 +0000

Sorry, by "official support" I mean that there is support within MSR to investigate these tools as a research project, no that there is a plan to actually do the conversion.

New comment by mkehrt in "Rust is tier-1 language at Microsoft"

mkehrt — Thu, 10 Sep 2026 18:57:20 +0000

As I pointed out in a sibling comment, the plan isn't for it to work. This is a job posting for a researcher at MSR to investigate what it might look like someday.

New comment by mkehrt in "Rust is tier-1 language at Microsoft"

mkehrt — Thu, 10 Sep 2026 18:56:12 +0000

It's less one guy's plan and more one Microsoft Research team's research goal to investigate technologies that might enable that in a few years. So probably more institutional support than just some guy, but less actually planning on succeeding in the full ambitious goal.

New comment by mkehrt in "Formalizing Fermat's Last Theorem"

mkehrt — Sat, 05 Sep 2026 18:40:45 +0000

I enjoyed this Terence Tao post the other day.

The relevant quote is

> one might naively expect that the natural question to ask with regards to a given problem X in a field is "What is the answer to X?". But in many cases the more valuable question is "What can be learned from studying X?"

And later

> But the currently fashionable practice of pointing a powerful AI tool at the task of answering a problem X, unguided by any human expert in the field X resides in, has created an unprecedented divergence between the production of answers, and the production of insight, to the point where the two questions have become _negatively correlated_:

https://mathstodon.xyz/@tao/117208618508728654

New comment by mkehrt in "I trained a small transformer in 1.5hrs and it beats many LLMs"

mkehrt — Tue, 01 Sep 2026 20:34:11 +0000

I don't understand what you're saying? An LLM is a transformer model trained on a large corpus of natural language, often with some post-training. An image model is a different type of transformer model. What's controversial here?

New comment by mkehrt in "I trained a small transformer in 1.5hrs and it beats many LLMs"

mkehrt — Tue, 01 Sep 2026 18:14:09 +0000

Sure, and those aren't LLMs?

New comment by mkehrt in "Rethinking Database Programming"

mkehrt — Tue, 18 Aug 2026 11:33:56 +0000

This isn’t talking about correctness of SQL. It’s talking about correctness of queries.

New comment by mkehrt in "A spectre is haunting Unicode"

mkehrt — Sun, 16 Aug 2026 00:24:39 +0000

So www.catb.org/~esr/jargon/html/M/mu.html

New comment by mkehrt in "“Code was never the hard part” is an insult to all programmers"

mkehrt — Sat, 08 Aug 2026 19:58:19 +0000

I think this is true when you realize "customer" means "coworker" or "person on the next team over" or "management" or "your intuition about the problem". Which all sort of have the same shape.

New comment by mkehrt in "Ancient Library – 1,060 Greek/Latin texts, click any word to parse it"

mkehrt — Sat, 08 Aug 2026 02:02:41 +0000

There are in fact rough breathing marks there on the articles, stacked under the inverted breve accents. Which is an alternate form of tilde.

New comment by mkehrt in "Ancient Library – 1,060 Greek/Latin texts, click any word to parse it"

mkehrt — Sat, 08 Aug 2026 01:58:12 +0000

Sort of. Romans didn't handwrite like that. See https://en.wikipedia.org/wiki/Roman_cursive . Note that the "u" character in both ancient and late ancient Roman cursive is more like the modern "u".

(Source: not an expert in any way)