squirrellous

Similarly, someone told me the memory consumption of a certain data structure couldn't be smaller because it had to store the full day's history of a thing rather than the snapshot state. We reduced it by about 30% in the end just by optimizing memory layouts and removing redundant data.

That confused me too since I remember reading they are Rust proponents. Apparently their storage engine TiKV is written in Rust, and TiDB is the database built on top.

That's more or less what happened with math, so I don't think it's an unfair expectation, especially with more AI advances.

One way I use skills, which I don’t see mentioned very often, is as “shortcuts”. Imagine some frequently issued prompt like “fetch origin and rebase this branch onto origin/master and resolve conflicts”. I make that into a little skills file called “rebase” with a one sentence description, and next time just type something like “/reb-tab-enter”.

Riegeli is great! Packed full of features. It could be the standard storage format for serialized pb binary if there were more official support.

Ah yes, something like prefix tree would work well storing reverse domain names like com.abc.www (although this is such a well known thing I feel like I must be missing things).

I wonder at their scale, why wouldn’t it make sense to store the entries lightly compressed in memory?

There is https://hnrss.github.io/ if you don’t mind a 3rd party service.

This doesn’t seem to mention that having AI aggressively optimize software will result in code that is too clever / complex to be maintained by the original human author. For something with a very stable API and can be tested to death like a regex engine it’s likely workable. Other things, not so much. We have enough trouble maintaining AI generated code _without_ aggressive optimizations as is!

I _think_ that has always been based on discretion of the mods. It’s only allowed after some period of time and if popular enough.

Well, did they hire you? Don’t leave us hanging!

Here are some that come to mind right now:

https://research.swtch.com/

https://danluu.com/

https://apenwarr.ca/log/

My first thought as well. Debugging is actually a decent use case of AI.

These days I am of the opinion that we don’t need a real IDL. Instead one should choose a “primary language” and define the schema in that language. All other language bindings are generated from that canonical language. Then for wire format you’d just pick something workable from pb json msgpack etc. This replaces all the Google codegen intermediate layer, and for the primary language it naturally becomes very ergonomic.

Honest question - why isn’t the following document [1] a standard? Is it too loosely specified?

[1] https://protobuf.dev/programming-guides/encoding/

Research done by SonarSource, a company that sells static analysis products. This explains the focus on static analysis, among other things.

Love to see someone implementing pathlib in c++. It is what stdfs could have been.

Those three things you mentioned kind of live in the same niche - offline data storage and querying. In that world yes everything has become columnar since it’s just better. Row-oriented is still the solution for online streaming use cases.

The last time I tried JJ it plainly refused to work with git LFS, which is kind of what OOP here is about (large files). Has that improved?

If you don’t like what Anthropic is doing, stop paying them money. There’s plenty of competition to go around. They can’t keep this up for long if users flock elsewhere.

One could argue the reinterpret_cast makes the intent more explicit which is a good thing.

That said I don’t have much against the use of void* or even char* here. If it works in C, it works in C++ just fine. std::span is not the right tool for this.

I think there are situations where AI is good for quick advice, provided you _also_ have a professional to talk to if needed. I sometimes seek advice about dealing with difficult colleagues, or ask for opinions about things I said at work while a little emotional. These don’t affect my personal life that much so the stakes are lower.

I am not very well versed here but I think due to the requirement for assembly and single-instruction commit, practical uses of rseq is generally very simple. It is nowhere near the usefulness of locks.

IIUC rseq is similar to thread-local data with the additional benefit that it scales with number of CPU cores, not threads. However if you are an application developer and is able to control all the threads in an application, then rseq isn’t that superior.

I fully agree that rseq should be more easily available to Linux developers, though.

This is the kind of comment I come to HN for. Thanks for teaching me something.

I’d guess that is what you get for training them on source code.

Also the I vs E axis feels meaningless in this case.

The point about feedback is spot on. However the generalization to system software seems a little sloppy. Something being system software doesn’t inherently make it more testable. You can have feedback-abundant and feedback-scarce pieces in both system and user-facing software.

Uhm what? All of those things are totally ordinary.

It’s shocking to me that you of all people should have financial issues. You are a legend in the community! By all means, take things closed source if it even just helps a little bit. As a profession we’ve all been hurt by over-sharing.

Both Facebook and Google internally use a custom version of mercurial, so I wouldn’t count it out just yet.