at this point, all i can think to do is make a job that frequently checks if it is running and starts it if it isn't

does service respawning in guix / shepherd only work when the service successfully starts once? or is it supposed to result in a retried start even if it fails the first time?

at this point i'm assuming the former...? the guix manual says shepherd "restarts the service when it stops" but i suppose it doesn't "stop" if it never started properly...?

i have a service that keeps trying to start too early. i couldn't figure out requirements that would work (advice for that would be great too), but at minimum i'd like it to retry starting. i never see that happen.

Status of wireguard-wg0:
  It is stopped (failing).
  It is enabled.
  Provides: wireguard-wg0
  Requires: networking user-processes sops-secrets NetworkManager
  Custom action: configuration
  Will be respawned.
  Failed to start at 10:09:39 AM (4 minutes ago).

it doesn't show, but the respawn-delay is 10, respawn-limit is (100 . 3) which should in practice result in no limit. and it's failing to start because it's failing to resolve hostnames because the system is not actually online yet.

#guix

starting to see the errors of my ways in my recent #guix config changes. there are multiple.

right now, one thing i am wondering is, do any of you split your guix config up into multiple files (modules), and if so, are those modules automatically part of the system generation so that using a previous generation will use an previous version of those modules, too? and if so, how do you inspect the previous modules? i only know how to find the previous configuration.scm

edit: it's magic?

edit: about provenance

is there an alternative client frontend to the official #fdroid app that has an anti-LLM policy? i really don't want something that is capable of installing and updating apps on my phone, of all things, to be slop-ridden; i don't care how minor the commits are or how well they've supposedly been reviewed, and i'm not going to act like it's pointless to try just because some android dependencies might have slop. that's just not how i see it. and i don't need the FSF to tell me "slop's okay actually" for me to change my mind. i can have my own opinions.

(i'm working on leaving the android ecosystem entirely, but that will take time.)

#NoAI

the wireguard service i've configured in #guix is not working because it always tries to start before my laptop is connected to the internet and fails to resolve hostnames. anyone know how i can make it wait until there is actual connectivity before trying? also how to make the service retry on failure at all?

it's already dependent on the networking service but that's apparently not enough. NetworkManager is generally what gets me online if it matters.

contrary to popular belief, the "e" in "email" stands for "evil"

my cat when i try to pet her after i've been petting other cats

this is very useful: https://codeberg.org/hako/emacs-guix-starter/src/commit/3b154d3ec165e0988ff2d5d40d4134d84f8c2e67/init.el#L49-L55

seems that those lines load the necessary dependencies!

ah, emacs-guix is a thing

seems that the search results that mentioned guix repl were onto something:

Compared to just launching the guile command, guix repl guarantees that all the Guix modules and all its dependencies are available in the search path.

- so says the guix manual.

,use (gnu home) has no errors in guix repl. someone had suggested making a script that launches guix repl and having that be the geiser binary. i tried it, maybe too hurriedly, and there were errors... but in realizing this, guess i'll try it again.

is this maybe not something people do?

do people just use the manual as a reference?

currently fumbling around, trying to figure out how to edit my guix-home-config.scm with emacs, geiser, and company for auto-complete. it would be *so nice* to have, but i don't know how to make it work. loading the buffer in geiser complains about "#~". trying to manually `,use` certain modules in the repl like (gnu home) gives me `no code for module (gcrypt hash)`

someone suggested adding "~/.config/guix/current/share" to the geiser-load-path, which i've done, but no dice.

#guix

i notice people talking about switching operating systems, and not just away from debian but away from linux entirely, with one popular destination being NetBSD in large part due to their strong anti-AI policy. seems to be a lot of enthusiasm at the moment!

to folks who have done this: was copyleft a consideration for you? the idea that the bsd license allows somebody to take the code, change it (or not change it), sell it, and have no obligation to release any source code - that gives me some pause. at the same time, it makes me wonder about the *actual usefulness* of copyleft guarantees in software today, and how useful they will be in the future...

but rather than get deeper into my own thoughts right now (which are not really settled), i'm curious about others'. if this was important to you prior to switching to a bsd, what changed?

#NetBSD

cats: what are they?

RE: https://mastodon.social/@eff/117215319335459732

EFF really sucks on this issue and they have for quite a while now, if you've been paying any attention. in a way, it's good to see them finally being more explicit about who they serve the most. surprise! the people come in second.

don't have heroes, folks

a proposal has appeared on F-Droid to adopt debian's AI policy (in an interim capacity, at least). if free and open-source software, android, and use of genAI have any particular significance to you, you may want to check it out.

https://gitlab.com/fdroid/admin/-/work_items/699

#FOSS #fdroid #f_droid #debian

interview with jose briones of dumbphones.org, by @techlore
https://www.youtube.com/watch?v=k2V4kuOJMR4

very interesting, especially as smartphones become more enshittified all the time

look, i don't want to bully #grapheneos. everyone be nice! but i'm tired of staying silent while their accounts spew unchecked lies for years, so i'll just say this and then go back to ignoring it again.

in my time working for #calyx / on #calyxos (a time which has now ended), despite graphene's claims to the contrary, i *never* saw anyone from calyx spread or encourage malice toward their project or community or harass them *at all*. if anything, due to a history of negative attention coming from their direction, calyx has long had a rule *not* to discuss their project in calyx's online spaces, but has still demonstrated respect for their OS in public, e.g. at conferences, acknowledging it's a good choice for some.

meanwhile, there has been a steady stream of smears, accusations, and now schadenfreude from their side about calyxos. you need look no further than their official social media accounts for evidence of this; you don't need the "receipts" from me to see the malice. it's all on full display, no background knowledge or rewriting of history required to understand.

contrary to their claims, during my time, by policy, calyx didn't even *look* at their code, let alone incorporate it (apart from some early remnants that are properly attributed, like signing scripts). in fact, with the exception of a weird, random proposal from the founder, as far as i know, calyx never even tried to interact with them whatsoever.

anyway... 4-5 years ago, before i had even finished deciding what OS to use, i was banned from grapheneos's chat channels before even getting around to send any messages - apparently just for existing in the calyx channels - and that made my OS decision easy: i wouldn't use an OS run by folks who'd do something like that. funnily enough, if not for that ban, and subsequently learning more about their behavior, who knows? i may have used grapheneos, just ignored the community stuff, and never ended up working for calyx in the first place.

since i know a lot of former #calyxos users are reasonably considering switching to #grapheneos (or already have), here are some tips for navigating their community:
* never question project leadership.
* never question security or privacy features or implementations.
* do not associate with any projects or individuals that they hate-stalk around the internet.
* no criticism, even constructive.
* make peace with the potential for baseless accusations and smears.

these tips also useful for oems. 💕

i know we all *want* to have an OS that is so secure and has such incredible isolation features that it's safe to install untrusted apps, as long as you don't grant them any special permissions. but i am sorry to say, unless you use a separate device for every untrusted app and keep that device turned off most of the time, you're kinda kidding yourself. sadly, regardless of how secure your OS claims to be, if you don't trust an app, you shouldn't install it, period.

link from last boost: https://arstechnica.com/security/2025/10/no-fix-yet-for-attack-that-lets-hackers-pluck-2fa-codes-from-android-phones/

#android #security #infosec #smartphones