Well to be fair, nix packages (as in the actual packages and not the modules/services) also mostly just extract archives. I mean the entire system is built on the assumption that 'config && make && make install' is how the thing is built.
Then nix layers way more nicety on top by offering modules, but nothing prevents you from ignoring all that and just using installing packages.
Brewfile is definitely the best option when in that ecosystem, but even that falls short by having no explicit versioning. You have to hack in version pinning if you run into an issue with a specific package.
If you like mise, check out devenv for the nix version of it. Its a lot of the nix benefits without a lot of the nix drawbacks (but still some)
Agreed on 1, thats part of the cost you pay when choosing Nix.
Disagree somewhat on 2. Yes, there are frequently small breaking changes, but the massive upside of nix is I can either just rollback or pin a specific package at a specific version and move on.
Nix plus flakes (and optionally devenv) is such a great baseline to operate agents in. So much less thinking about dependencies on various runtime platforms, and you get pinning out of the box.
My permissions prompt isnt quite working right with it either. It pops up but isnt blocking, so claude continues editing and asking for other permissions which replaces this prompt. Then when you confirm those prompts, it shows the LSP prompt again. Definitely needs polish (and explanations on how it even benefits the agent)
Yeah, I posted here because I was completely blindsided when my claude asked if I wanted to install a go lsp. I didnt even know that was a thing. A little googling led to this changelog from 3 days ago, but I was surprised I hadnt seen any previous mentions of this online (from either creators, anthropic, or HN posts).
I am disabling it for now since my flow is fine at the moment, I'll let others validate the usefulness first.
I had a conversation with Claude code 2 weeks ago where it mentioned early support for LSP had been added into Claude code. Have been working on a LSP for a custom language since then.
In general anything with an API is simply saying "find the auth token at ~/.config/foo.json". It mostly knows the rest endpoints and can figure out the rest
That has not been my experience with sonnet, and even so it is largely remedied by having better AI docs caching the results of that investigation for future use.
How would you evaluate it if the agent were not a fuzzy logic machine?
The issue isnt the LLM, its that verification is actually the hard part. In any case, its typically called “evals” and you can probably craft a test harness to evaluate these if you think about it hard enough
Then nix layers way more nicety on top by offering modules, but nothing prevents you from ignoring all that and just using installing packages.
reply