Petr Viktorin

@arfisk@aus.social They just... um, found it! Right? It was buried, in a ... potato field! They were digging up potatoes as a side quest, you see, and found treasure. Very lucky, right? https://www.peppercarrot.com/en/miniFantasyTheater/055.html

@itamarst File a bug for stdlib? Meanwhile you can use the cgroups2 CPU bandwidth API: read two numbers from /sys/fs/cgroup/cpu.max, divide, round up. https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html#:~:text=cpu.max

@christophehenry@mastodon.social Not necessarily; if you learn assembly, C++, or Rust, you'll get the concepts as well. But also, in code where locality or SIMD matter, C will be extremely similar to C++ or (unsafe) Rust. And you'll be staring at assembly when optimizing it, regardless of language. No need to learn quirks of the C preprocessor or dark corners of libc, of course -- other languages are better at that, including Python. But "learning C == learning pointers" is a reasonable simplification.

@nedbat@hachyderm.io @christophehenry@mastodon.social You don't need to understand stack vs. heap to make a web app, no. But you probably do need it before diving into SIMD. And C is still the best tool for *learning* stack vs. heap. Mostly because it lets you make catastrophic mistake easily, rather than give you tools to prevent them (C++) or require those tools (Rust).

@nedbat@hachyderm.io @christophehenry@mastodon.social Hmm, the article points out the enormous effort CPU designers put into maintaining the interface (assembly). You need to know pointers and stack vs. heap, before you go exploiting implementation detailsm (e.g. avoiding pointer chasing) -- like you needed to know function calls in Python2 before starting to avoid them for efficiency.

@b0rk@social.jvns.ca This makes the third way look like an internal detail, but to me, "current state of the branch" is useful. git show BRANCHNAME or a GitHub .../BRANCHNAME/FILENAME URL look at a branch this way.