Without thinking about it much, my understanding was that each line of the stack trace referred to a real line, even though the block as a whole wasn’t a program.
But! because of this comment I went and checked the lines of those stack traces. And in fact, they’re not real lines, just the C++ type expansion.
That said I’ve got a another half as bad example that is real so Ive edited the comment to point to that example instead.
If you think that’s good, then you’re gonna love this “simplified” real code posted as a real issue on one of my Github repos.
Edit: updated link to address the stack-trace comment
I’ve been tracking the version search issue on nixpkgs for the last 3 years. Your site is a major win on that front!
I still use Lazamar’s tool regularly, so I’m thrilled to see a successor to it.
Devbox (a framework) has the most exhaustive version search for nix that I know of, but it’s a bit of a pain because they assume you use devbox to install the package (instead of using nix to install the package)
Can we get the links/receipts/examples?
I’m onboard with the post, and glad to see Lemmy working as intended. Making a post (instead of silently defederating) is wonderful. That said, I think its very important to say “because of comments like X[link] and Y[link]” instead of “due to severity”.
Imagine if the roles were reversed and the CCP was saying they’re “defederating from lemmy.world due to severity”. Members should be given the 1st hand evidence, rather than 2nd hand opinions/summaries.
Man this is the best thing to happen to nix since it became mainstream.
I’ll mention some things others haven’t;
Usually, like if a python package isn’t available, you can do half-nix by using something like python venv, and install everything like normal in venv. Or for rust, you get rustup through nix, but then let rustup do its mutable work, instead of literally managing crates through nix. It’s also possible to patch a lot of stuff into a nix shell for inter-op.
When that fails, which it does, then its
(I know I’m two months late)
To back up what you’re saying, I work with ML, and the guy next to me does ML for traffic signal controllers. He basically established the benchmark for traffic signal simulators for reinforcement learning.
Nothing works. All of the cutting edge reinforment algorithms, all the existing publications, some of which train for months, all perform worse than “fixed policy” controllers. The issue isn’t the brains of the system, its the fact that stoplights are fricken blind to what is happing.