brian-armstrong

Everyone is highly sensitive to slop, and it still looks like slop.

If you actually want something that looks decent, it feels like tool use would at least help cover the obvious blunders. Building layers to composite would go a long way to removing some slop here.

I'm still holding out for the sequel

Oh no! This is going in the wrong direction, isn't it? From the headline, I'd assumed the takeaway was encouraging small changes (and 1000 LoC is not "small" by any definition) and maybe encouraging the author to refactor at larger sizes to make human review easier. Instead it's just bypassing human review entirely?

Code review is such a key step to the feedback loop and helps keep people current on the changes as they come out. Skipping that entirely is just inviting disaster.

Thank you, glad you liked it

Ah, in that case, you would certainly be excused for trying to kill one or two of them.

The people who still use Windows yet complain about this are quite confusing to me. There's been every indication that it will keep going like this for years now, and the cost of switching is just not that high. If you're still riding that train it's time to just embrace it.

Is this related to the streaming service of the same name?

Interesting. I'll have to give it a try. If I understand your suggestion, you're saying that rather than doing 2 lookups for logs, then an addition, and then another lookup to exp back, we could just precompute the whole thing for any 2 arguments and do one lookup in a table? It's worth noting that some of the operations already use the mul_log format which just does an add with wraparound, but I could certainly see the big table option helping when we truly need field.mul

As far as LDPC, I'd love to, but I need to get a better understanding of them first. I'll probably go for Turbo next just because conceptually they feel closer to the convolutional codes I have already. But time willing, it'd be fun to do both.

Oh! Gotcha! I put a table in the README that lays it all out https://github.com/brian-armstrong/fec#performance

tldr: My library matches or beats libfec's SSE2 assembly for convolutional codes and pretty steadily beats it in Reed-Solomon. For the convolutional codes, my crate is using a generic, templated decoder rather than hand-written assembly, so it was nice to see that I could match the performance.

I used the benchmarking binaries that ship with libfec. My own crate has a libfec-compatible C shim, so I can link the benchmark against the Rust crate. The benchmark itself reports time spent for a given number of iterations, so the throughput can just be extrapolated from that.

You could certainly call this crate through a shim in goestools but you'd have to add a Rust dependency. But you could also rewrite goestools in Rust, though it wouldn't be a small undertaking. This crate does have everything you'd need for the forward error correction, at least.

The models would have to get significantly better for this to help, though. Unedited LLM dialog is quite bad. Though, I wouldn't put it past AAAs to try anyway.

I read these stories and I can never figure out how people are managing to use these $200 plans. If I really go full bore, I can sometimes max out the $20 plan. Even then, it already produces more code than I can reasonably review and merge.

Absolutely. Let them fend for themselves in the streets.

Yikes. That's a car that looks like it gets its lunch money taken by the other cars.

It's good, but you still have to pay monthly for it. Feels like it kind of defeats the point of having a local collection.

What? Most (all?) platforms let you opt out of this, and you receive payment when your shares are borrowed. And shares are fungible, how would it matter if one share is borrowed by 10 people or by none?

Not comparable at all. Xbox would be mostly transient traffic. It's probably not much more than packet forwarding for a lot of traffic.

Github is a giant complicated stateful mess with a lot of reads and writes. It also has a lot of features at this point. Hard to scale and hard to optimize.

Why would you ever accept a mismatched certificate? Even assuming that you think your ISP has no nefarious plans, are you going to be able to rigorously confirm it's their certificate? At that point you've bypassed all the mechanisms in your browser that do this heavy lifting for you.

Or the Cinco Encyclopedia of Numbers https://youtube.com/watch?v=ZI3QxgjLGbQ

Thanks, Cinco

It'd be nice if AI-written articles had an [AI] after the headline

What kind of self-disrespecting dev is using MacOS in TYOOL 2026?

It seems unlikely that "is user adult" is not already easily modeled by any of these companies to within a very high degree of confidence. Even 15 or 20 years ago Google search could bracket your age pretty effectively. It doesn't seem like this adds metadata that wasn't already there.

Location: SF / Remote

Remote: Mostly (Open to 0 - 3 days/wk in office)

Willing to relocate: No

Technologies: Rust, Elixir, Typescript, Python

Linkedin: https://www.linkedin.com/in/brian-armstrong-sf/

Email: brian.hn.hireme@gmail.com

Hello HN! I'm a Staff Software Engineer with a variety of backend experience. I helped Discord scale its WebRTC system and helped guide its media infra (blog post: https://discord.com/blog/how-discord-resizes-150-million-ima... ). Currently open to work where I can put my skills to use and help get your projects into production.

Some of my personal and working traits: I'm a big fan of data-driven decision making. I think the best time to think about "how are we going to deploy this" is before any code gets written. I think empathy is underrated and I prefer over-communication to under-.

Cheers!

You don't really need Windows for gaming anymore unless you're playing the games that absolutely insist on kernel-level anti-cheat. Proton is extremely good on Linux these days.

Location: SF / Remote

Remote: Mostly (Open to 0 - 3 days/wk in office)

Willing to relocate: No

Technologies: Rust, Elixir, Typescript, Python

Linkedin: https://www.linkedin.com/in/brian-armstrong-sf/

Email: brian.hn.hireme@gmail.com

Hello HN! I'm a Staff Software Engineer with a variety of backend experience. I helped Discord scale its WebRTC system and helped guide its media infra (blog post: https://discord.com/blog/how-discord-resizes-150-million-ima... ). Currently open to work where I can put my skills to use and help get your projects into production.

Some of my personal and working traits: I'm a big fan of data-driven decision making. I think the best time to think about "how are we going to deploy this" is before any code gets written. I think empathy is underrated and I prefer over-communication to under-.

Cheers!

I'm curious, how did you make it this far in life without realizing that paint is delicate and scratches easily? Do you have untreated brick walls in your house or something?

It would surely depend on the SSD and the firmware it's running. I don't think you can entirely count on it. Even if it were working perfectly, and your strategy was to power the SSD on periodicially to refresh the cells, how would you know when it had finished?

It's quite possible. Some SSDs are worse offenders for this than others. I have some Samsung 870 EVOs that lost data the way you described. Samsung knew about the issue and quietly swept it under the rug with a firmware update, but once the data was lost, it was gone for good.