systemd-jaded.timer

purely dysfunctional programmer, soft and snuggly version; they/she.

Last Notes

2026-01-11 05:07:28 UTC

OH: so basically girls need to teach me category theory because it is hot

2025-11-24 07:36:53 UTC

OH: "I think #Lean is cool because it's like they trapped a graduate student in the compiler and they just prove your array accesses are in bound for you"
@nprofile…af8j blink twice if you are trapped in a compiler

2025-10-26 22:42:39 UTC

OH: baseball is just eve online for sports fans

2025-07-15 07:37:39 UTC

- reply

@nprofile…vklh @nprofile…cds8 i guess? but Tailscale is always on and from the perspective of the programs using it, it's zero configuration; obviously not so from the perspective of the system as a whole. but nor is mTLS. mTLS is theoretically pretty good in this regard because you can shove a client cert into a system certificate store and even many CLIs work (insert many asterisks here), unless you're on Linux in which case those don't exist.
so the amount of frustration you have to deal with to build on top of it as a platform at least on the client side is, with my experience with very limited dealing with mTLS and its utterly inconsistent implementation between applications (*cough* postgres), vastly greater compared to Tailscale.
it's approximately the same reason i wrote Clipper: i don't want to deal with applications' bespoke tls certificate or proxy configurations, i just want the keys in a uniform manner so i can just read the traffic. same with Tailscale: it just works and you don't have to worry about bespoke app configuration.

2025-07-06 17:35:46 UTC

- reply

@nprofile…7dgh FASCINATING. i love hearing about people's workflows, especially if they managed to combine together the right advanced features to get what they want (and what they want is a creative use of the tool).
i think originally i used symlinks (and my own bad script) because i didn't like the aesthetics of any of the dotfiles managers which copy the structure of the home directory and didn't want to learn yet another thing. these days, in principle, I've become an advanced enough git user that i could have adopted your workflow but i still have incompatible opinions about the layout that lead to using symlinks.
you could definitely write a blog post about this invention.

2025-07-06 17:14:53 UTC

- reply

@nprofile…7dgh there's a horrible solution to this actually. you can put * in gitignore and then !foo for each thing you actually want. i know people who do this weird git pattern and different people who put their home directory directly in git.
i don't personally think the latter is great because i don't like the home directory layout and think it's poor code organization. also it will make it pretty easy to do weird tooling mistakes with things that expect a parent repo. but it is technically possible!

2025-07-06 03:07:00 UTC

new post about how you can choose not to use #Nix to manage some parts of your configuration, how to do it, and the benefits you get from doing that
instant changes, no rebuilds, and you can still get pinning for dependencies
https://jade.fyi/blog/use-nix-less/

2025-06-10 20:44:27 UTC

- reply

@nprofile…6m2n CPPFLAGS is for the preprocessor, though I guess it's strictly correct to use instead of CFLAGS for -I in particular! I just think it might be confusing to refer to CPPFLAGS rather than the more common CFLAGS/CXXFLAGS.
I think one confusing binary toolchain thing that got me when I first learned is that I didn't understand that cc was both the c compiler *and* the linker depending on args, and that often tooling would be really confusing about what it actually is calling when it is "calling the linker" (-fuse-ld referring to an actual linker) *because* a lot of the time, including with rust, "calling the linker" means "calling the compiler driver in link mode".
source: https://docs.jade.fyi/gnu/make.html#index-CPPFLAGS

2025-04-09 01:49:51 UTC

your occasional reminder that (( "$foo" > 5 )) and -eq in [[ ]] is a command injection bug in #bash
https://yossarian.net/til/post/some-surprising-code-execution-sources-in-bash

2025-04-08 20:37:15 UTC

- reply

@nprofile…6m2n i have some truly heinous contrast bugs with solarized dark and diff tools like nix-diff. idk what the fix is besides moving the offending colour far away from what it currently is on a case by case basis and doing a lot of "hoping". i think the way it's bonus fun is that i use nvim with termguicolors and just the default terminal theme and the problem colours might be getting passed through or something (though I'm not confident, i need to stare at it properly).
to some extent the problem is that the tools in the terminal don't have any idea whether the primary colour has acceptable contrast with any "red"/"yellow"/etc.

2025-03-29 03:33:58 UTC

omg rust is woke. it's good actually. https://diziet.dreamwidth.org/19480.html

2025-03-22 01:19:46 UTC

man. what a day of mostly just doing code gardening. it needed gardening, just, i sure did a lot of it. filing bug reports, merging changes, etc.
but this *is* how you get stuff to be better, eventually.

2025-03-16 07:48:18 UTC

it turns out that pee is really useful! you can `pee cat logger` to log stuff to both syslog and the standard output, if you want to diagnose something going awry without being able to see the output directly (but don't want to suppress its normal destination)
STOP LAUGHING OK
#linux #pee

2024-01-02 11:54:47 UTC

Nix flakes aren't real and can't hurt you: how to write flakes the non-flake way.
This post is about the macro-level composition primitives one wants while writing nontrivial Nix software, and how to write packaging that is as reusable as possible: cross compilation, clean portability between projects, extensibility.
It also briefly goes over why one might use the six different ways of getting software with Nix.
https://jade.fyi/blog/flakes-arent-real/