das_j has quit [Quit: Bridge terminating on SIGTERM]
das_j has joined #nixos-chat
<gchristensen>
veleiro: you have a great quit message
<gchristensen>
oops. veleiro
<gchristensen>
.. ??? V
<V>
gchristensen: thanks :D
<V>
You're the third person to compliment me on that tonight
supersandro2000 has quit [Disconnected by services]
supersandro20000 has joined #nixos-chat
srk has joined #nixos-chat
<infinisil>
Arghh, for some reason, when I cd to a directory that has a stack.yaml file in it, it tries to evaluate some Nix Haskell shenanigans
<infinisil>
And I have no idea what's causing that
<infinisil>
It's not direnv. It's not the chpwd function from zsh
<infinisil>
But also, it doesn't happen all the time
<sterni>
o_O
<infinisil>
i guess if I manage to set -x before it happens i might get a clue (though zsh's trace is very verbose if you have any kind of customization)
<infinisil>
Oh wow, I think it's an interaction between powerlevel10k and stack's awful nix integration
<infinisil>
p10k detects there's a stack.yaml, it wants to show the stack resolver version in the prompt
<infinisil>
So it calls stack with some arguments
<infinisil>
Which on NixOS attempts to just evaluate something like `(import <nixpkgs> {}).haskell.packages.ghc<version in stack file>`
<elvishjerricco>
infinisil: Yea stack's nix integration is pretty annoyingly designed
abathur has joined #nixos-chat
<abathur>
anyone feel like they have a handle on cross-platform C locales? :)
rajivr has joined #nixos-chat
patagonicus2 has joined #nixos-chat
patagonicus has quit [Ping timeout: 272 seconds]
patagonicus2 is now known as patagonicus
<infinisil>
Will email ever disappear?
<infinisil>
I mean, it will eventually, but when?
<worldofpeace>
it could be arranged 🔫
<infinisil>
Heh
<infinisil>
I'm thinking that in 50 years email might not be a thing anymore
<samueldr>
abathur: no
<samueldr>
I'm answering for everyone
<abathur>
<3 samueldr that's certainly how I feel about it
<{^_^}>
samueldr's karma got increased to 324
<samueldr>
ugh! I just painted myself into a corner
<samueldr>
pun almost intended
<infinisil>
Hm though 50 years is not a lot, I'll go for 100 instead
<samueldr>
I started to label my phones with the codename on the display-side, putting them in a corner on the bezel
<samueldr>
until I got to one with a teardrop camera, and no real bezel
* abathur
watches people with no-bezel phones try to use them without accidentally interacting the edges
<samueldr>
don't worry
<samueldr>
they're using AI to solve the issue
<samueldr>
(not really... but almost)
<samueldr>
in the next android android release, apparently they'll "solve" the full-screen back gesture, which interferes with side-drawers in apps, with AI
<samueldr>
"AI"
<abathur>
uhuh
<samueldr>
so by my estimates it will mean that whatever worked maybe will work, maybe won
<samueldr>
won't next time you try it
<samueldr>
meanwhile, I have the solution: put the navbar back
<abathur>
I talk to OK-Google a lot so I'm pretty skeptical :]
<samueldr>
or go back to the drawing board and do a playbook ok
<samueldr>
where the **bezels** (which have disappared these days!) had touch sensitivity
<samueldr>
so playbook OS could differentiate from a drag from the top of the screen, compared to a drag from outside of the screen entering the top
<samueldr>
the experience of a playbook OS tablet was really out of this world at the time it was released
<samueldr>
(too bad their hardware has a secure boot chain that hasn't been broken)
<abathur>
infinisil: anyways, idk, it seems to be mostly driven by the good-enough replacement
<abathur>
it's taken a really long time for digital to beat paper for ~official communications from many corp/gov orgs
<abathur>
landline -> cell transition was fast, but driven by interop of tech and addressing
<abathur>
imagine how much longer it could have taken to treat your cell as your primary number if there wasn't landline/cell interop
<samueldr>
my guess would be about 1.5 to 2 × the lifetime of fax machines until they are all displaced
<abathur>
driven wrong word; enabled by I guess, not resisted by
<samueldr>
>> He received British patent 9745 on May 27, 1843 for his "Electric Printing Telegraph"
<samueldr>
so I guess it'll be a while
<samueldr>
so 260 to 350 years, if we were able to displace all fax machines this year
<abathur>
aim and MSN and gchat and hangouts and snapchat and facebook messenger and whatsapp and signal can all displace rapidly because they still aren't load-bearing in the way one's interface with large bureaucracies is; 100 seems like the better bet, as weird as that is to say
<abathur>
hopefully email's 2nd century is better than its first...
<samueldr>
if doctors and pharmacists here are still using 1843 technology for official communication
<samueldr>
I wouldn't bet on e-mail being displaced that quickly as only in 100 years
BaughnLogBot has quit [Ping timeout: 272 seconds]
BaughnLogBot has joined #nixos-chat
aszlig_ is now known as aszlig
<aleph->
elvishjerricco: So went and read your post on nixops key deployments and pass, finally went and implemented a tweak of it too. Works rather nicely.
<elvishjerricco>
aleph-: It's ok. I still use it, but I'm not hugely happy with it. For one, NixOps is just very slow and kind of annoying to use. But it really annoys me that it doesn't delete obsolete keys
<elvishjerricco>
How'd you tweak it?
<cole-h>
elvishjerricco: Maybe you could apply some sort of tweak yourself by following what sops-nix does: creates /run/secrets.d as a ramfs, creates a new generation on every switch (e.g. /run/secrets.d/1) and then symlinks that to /run/secrets
<cole-h>
Then you could probably extend it to remove prior generations on switch
<{^_^}>
ryantm/agenix#27 (by cole-h, 19 hours ago, open): modules/age: symlink files into place
<cole-h>
(though, sans the "remove prior generations" part)
<aleph->
elvishjerricco: Eh to be honest just went with an inline call of `nix-instantiate --eval -E "builtins.readFile < $(pass $1)"` roughly from memory
<aleph->
No need to write to a temp file and TRAP and rm on exit
<aleph->
Hmm think there was something different. But it was roughly that
<aleph->
elvishjerricco: Is nixops slow? Found it to be decently performance.
<aleph->
performant*
<aleph->
But then I have a lower host count
<elvishjerricco>
aleph-: Well my config for just one machine is already noticably slow to eval, and nixops evals the whole thing like 3 separate times for each deploy
<aleph->
Huh weird
<Ke>
yeah definitely nix being slow is an issue, especially on my aarch64 systems that have moderately slow single core perf
<Ke>
though a lot faster than portage
JJJollyjim has joined #nixos-chat
JJJollyjim is now known as Guest67960
<Ke>
obviously this is only an issue, if you are creating new configs, otherwise I don't mind letting it just run on the background
<Guest67960>
i feel like nix is an order of magnitude slower than it could be
<Guest67960>
but several things scare me off trying to help with that
Guest67960 has quit [Quit: authenticating]
Guest67960 has joined #nixos-chat
Guest67960 has quit [Client Quit]
Guest67960 has joined #nixos-chat
<Ke>
one of the things that I always wondered was, why are all modules evaluated and not imported conditionally instead of using service. .enable
<Ke>
like it could be modules = ["sshd"];
<Ke>
or imports = with pkgs.modules; [sshd];
<Ke>
but maybe that does not take any considerable time?
<Ke>
I use this approach for all my custom modules, they only get imported, where enabled and automatically get enabled when imported
<ashkitten>
Guest67960: btw since you're on matrix you might not be aware that your irc nick is Guest67960
<ashkitten>
how do i type it out without matrix translating it
<ashkitten>
it's Guest 67960 without the space lol
<etu>
ashkitten: Maybe with backticks around it
<etu>
Since they do code blocks
<sphalerite_>
`Guest67960`
<LinuxHackerman>
yup works
<LinuxHackerman>
though it doesn't show as inline code
sphalerite_ is now known as sphalerite
<sphalerite>
ah, the joys of chat bridging.
<ashkitten>
that's the thing that's always funny to me about matrix users on irc, is they never know what their irc nick is
Guest79985 has joined #nixos-chat
<Guest79985>
ty ashkitten, i think i should be both?
<Guest67960>
[test]
<Guest67960>
yeah i think i'm both
<f0x>
ashkitten: i can see matrix.org bridge users irc nicks from matrix, does that count :P
<Guest67960>
(I moved to a new matrix homeserver
<Guest67960>
but I can't talk to the appservice on the old server due to Reasons, so i had to change my nickserv password so it can't authenticate anymore
Guest67960 has quit [Quit: authenticating]
Guest67960 has joined #nixos-chat
Guest67960 is now known as JJJollyjim
JJJollyjim has quit [Client Quit]
JJJollyjim has joined #nixos-chat
JJJollyjim is now known as Guest26650
Guest26650 has quit [Client Quit]
Guest26650 has joined #nixos-chat
Guest26650 has quit [Client Quit]
Guest26650 has joined #nixos-chat
Guest26650 has quit [Client Quit]
<Guest79985>
ugh i think i broke something on the new one
<Guest79985>
i want federated irc bridging
<Guest79985>
matrix pls :(
<f0x>
what is broken?
<Guest79985>
neither nickserv nor appservice-irc will talk to me, no replies to my messages
<Guest79985>
(sorry if this gets doubled-up, i sent it on the new one and it didn't seem to go through)
<f0x>
are they e2e dms?
<f0x>
if not it might also be worth a try to just try starting a new dm anyways, i've seen someone have a similar problem before
<Guest79985>
the DMs only just broke, I'm guessing it's some sort of rate-limiting
<Guest79985>
oh hey you must be the f0x from the federation tester haha
<f0x>
*frontend, yes :')
<f0x>
my claim to matrix fame
<Guest79985>
i love the animation around the table on your website lol
<Guest79985>
hmm how do I start a new dm? If I just pick that user in the New DM ui, element takes me to the old one
<Guest79985>
and the appservice doesn't have a DM equivalent to !join
<Guest79985>
(I can now talk to appservice but not nickserv)
<f0x>
i'd create new room, not-public but also disable e2e, then invite just the nickserv ghost there
<Guest79985>
hmm, how does the bridge know to make that an irc DM rather than IRC invite?
<Guest79985>
(doesn't seem to work anyway)
<joepie91>
dirty little secret: no such thing as "DM" in Matrix
<joepie91>
:P
<Guest79985>
yeah exactly
<Guest79985>
but there is in IRC
<Guest79985>
that's the problem
<Guest79985>
how does the bridge know what to do?
<joepie91>
not sure how it works internally exactly, but one thing is that it could distinguish between "we're in a bridge-controlled room" as being an IRC channel, vs. "bridge-controlled user is invited to a non-bridge-controlled room" which constitutes a DM
<Guest79985>
hmm yeah no idea what's going on here now
<Guest79985>
ah
<Guest79985>
appservice tells me I am not connected to any networks
<Guest79985>
I can see that being a problem
__monty__ has joined #nixos-chat
<Guest79985>
guessing my bridge's ipv6 address is banned from freenode for too many bad password attempts, or something of that nature
<Guest79985>
I wish I had any visibility into that...
<f0x>
ah you're running your own bridge?
<Guest79985>
no
<Guest79985>
but I believe internally, matrix-appservice-irc has a different ipv6 address for each user on each server
<arahael>
Guest79985: You sould always ask #freenode about it.
<Guest79985>
so freenode-side, IP bans can be used
<arahael>
*could
<f0x>
yeah, appservice-irc should complain about that in the dm though
<Guest79985>
yeah it doesn't :/
<f0x>
are you in any bridged rooms with that account?
<Guest79985>
it looks like it, matrix-side
<Guest79985>
but my messages don't send
<Guest79985>
will ask in #irc:matrix.org, I'm sure freenode people are sick of dealing with matrix shit lol
<joepie91>
Guest79985: fwiw, I'd be curious to hear what the issue was, this is not a bug I've seen before
<joepie91>
when you figure it out
<Guest79985>
hmm something cursed is happening here
<Guest79985>
when you mentioned 79985
<Guest79985>
oh wait nvm
<Guest79985>
yeah the bridge translated that to an @ of my old server
<Guest79985>
but element also notified me about it on the new server, because it contains the substring j a m i e
risson_ is now known as risson
<LinuxHackerman>
Is there a community overlay (or non-overlay packaging) for pgtk emacs? I've been dragging around my own file which I think I originally stole from Graham, but I update it far too rarely…
<adisbladis>
LinuxHackerman: Yeah, we have that in emacs-overlay
<f0x>
good thing I have my nixos stuff in a separate tab group :P
<joepie91>
them are rookie numbers
<f0x>
i close things sometimes, yeah
<joepie91>
:P
<hexa->
f0x: morph :D
<hexa->
loving
<hexa->
the context
<f0x>
ah yes, morph is still a while away in this setup
<f0x>
I want to get CONFIG_USB_SERIAL_CONSOLE kernel flag working but no success yet
<f0x>
to get boot output over my CP2104 usb to serial adaptors, it's still unclear if that would even be supported, unfortunately
rajivr has quit [Quit: Connection closed for inactivity]
endformationage has joined #nixos-chat
sterni has quit [Ping timeout: 260 seconds]
sterni has joined #nixos-chat
<f0x>
I just tried nixos-enter, but doing nixos-rebuild switch inside that gives an error: writing to file: Operation not permitted for nix store paths?
<f0x>
i guess just running nixos-install again is easier?
<srk>
yep
<ldlework>
I just heard "NixOS is the Rust of operating systems"
<joepie91>
f0x: just running nixos-install again is what I do as well if I somehow manage to mess up eg. the bootloader
<gchristensen>
neat
<f0x>
joepie91: yeah mostly just odd that the rebuild didn't work
<ldlework>
gchristensen: insane learning curve, inhuman tooling, bizarre error messages, but solves some long standing problems in sound ways, and so willing to adopt early to trade those bigger problems for smaller ones while corners get rounded
<gchristensen>
Rust has fantastic error messages though and Nix does not :)
<ldlework>
One idea regarding nix errors I had the other day is when you are rebuilding your system and have paralleism enabled, the messages all get interleaved
<ldlework>
But if they were prefixed with a unique id, you could copy the output into your editor and easily filter for the messages of a given build thread or whatever
<samueldr>
those would be build-time error messages, right?
aaronjanse has joined #nixos-chat
<samueldr>
kind of one step too far for nix itself
<ldlework>
?
<samueldr>
it's not in the nix eval
<samueldr>
and prefixing all messages from stdout/stderr is not a trivial thing to do
<ldlework>
certainly nix controls the parallelism of building multiple derivations at once
<ldlework>
as it's a nix-level option of whether to build with paralellism
<ldlework>
i'm not talking about a single individual package using parellelism to build
<ldlework>
i'm talking about nixos-rebuild
<samueldr>
I know
<gchristensen>
I didn't mean to refer to builds but nix's errors
<ldlework>
I know this is just a separate thought concerning readability of nix tooling output
<samueldr>
what could be better is to better list all the actual .drv paths that failed at the momeny nix exits
<samueldr>
so then you can `nix log /nix/store/eeeee-that.drv"
<samueldr>
probably finishing up with "HINT: use `nix log [drv]` to get the build log for a derivation"
<srk>
f0x: yep
<gchristensen>
yea
<srk>
f0x: consult bootctl
<f0x>
right, second question, WHYY :P
<ldlework>
samueldr: what srk just said is what i was getting at, fwiw
<samueldr>
I really hate having to fish for the right `builder for ... failed with exit code 1`
<samueldr>
ah, I assumed the protocol had them already combined, as I believe I was told that at one point, or understood that
<ldlework>
if at some point nix has the independent output streams, it could prefix them for easy filtering. not saying it's the way to go, but it doesn't seem like the technical impossibility
averell has quit [Remote host closed the connection]
<ldlework>
ah yeah fair enough :)
<cole-h>
nixUnstable does this
<cole-h>
nix build -L
<samueldr>
still, getting the start of a line for a stream is not trivial, even with one
<f0x>
i wanted to switch to UEFI grub so I could set some other options, but that doesn't seem to have worked
<samueldr>
(but much more feasible)
<cole-h>
each log line is prefixed with the derivation's name
<ldlework>
nice, i predicted the present
<f0x>
hm i wonder if it's doing stuff in the wrong dir, /boot vs /boot/EFI
<cole-h>
Whee, solo v2 reached the TOTP stretch goal!
<bbigras>
It's a bit silly to lock that feature behind a goal. I mean they would make their product less competitive just to raise more money?
<bbigras>
maybe it cost more to get it but still.
<bbigras>
was some goal not reached?
<__monty__>
Do other yubikeys support TOTP?
<bbigras>
I would be pissed if some new elliptic curve keys were not supported.
<bbigras>
I have no idea. to be honest I don't know if I need TOTP. it's more the principle. like if they locked ed25519 behind a goal I would be pissed.
<bbigras>
I still ordered one when someone here first mentioned the kickstarter. but I want every single cool feature in my toys.
<__monty__>
Then why back it before it has reached all the stretch goals?
<bbigras>
I wonder if those goals were announced when I backed it. probably. I just assumed it would be feature complete. Why release a product like that if it doesn't have everything people expect from them.
<bbigras>
I expect it to be on par of the yubikeys but open
<__monty__>
You seem to be arguing the stretch goals should be silly things like cosmetics or something. While those are probably cheaper to come up with than the technological challenges.
<__monty__>
Having interesting stretch goals is a cheap way of getting people to advertise your campaign.
<bbigras>
I think I would prefer if their initial goal was higher, but the product would have everything.
<__monty__>
And it's not as if they can't release what's promised and then iterate on the product before they start selling it.
<__monty__>
What you prefer and what results in raising the greatest capital aren't likely to align.
<bbigras>
duh
<bbigras>
if the stretch goals are all software and they can be added later by them or PR, then my complain is invalid.
<__monty__>
Even if they require hardware changes, they can just do a small run to satisfy the kickstarter conditions. Then iterate on the design for mass production. *With* partial beta testing.
<bbigras>
it would suck for the backers having the inferior version.
aleph- has quit [Quit: WeeChat info:version]
<joepie91>
ar: neat trick
<__monty__>
Yup, just as it sucks for all the backers who've backed projects that failed to produce anything at all or produced inferior products. Or the products they promised but so late they'd become irrelevant.
<bbigras>
in this case I think the hardware was already finished. it says "ready to ship" and that they already have 1 batch. but fair enough.
<__monty__>
Note that I'm not saying this is the best outcome. It's just the local "optimum" you get in the capitalist system we have.
<bbigras>
yeah
<abathur>
turns out, "secret" is one of those words that starts to look weird if I read it too much
aleph- has joined #nixos-chat
<__monty__>
Semantic satiation, happens with lots of words.
<gchristensen>
maybe even every word
<gchristensen>
imo if you're experiencing semantic satiation your abstraction isn't quite right
<abathur>
it does; TIL there's a term, though; <3 __monty__
<{^_^}>
__monty__'s karma got increased to 0b11110
cole-h has quit [Ping timeout: 265 seconds]
<abathur>
idk; I wouldn't doubt someone can experience it with all words but my subjective experience is that some words are much more prone than others
<abathur>
but in this case I really just meant that the spelling starts to look wrong the more I read it :)
<__monty__>
What abstraction is at play here? Especially in the case of this happening in your native language?
<__monty__>
gchristensen: ^
<__monty__>
bbigras: Oh, one way the stretch goals could actually make sense is if they have a killer feature which can outweigh the lack of some convenience features like TOTP.
<__monty__>
"Secure upgrade" in the case of SOLO V2 for example.
<gchristensen>
I've only ever experienced it programming
<__monty__>
#humblebrag : )
<bbigras>
__monty__: do you mean like a better firmware that they could do anyway years later or maybe in the next kickstarter in case the stretch goal wouldn't be reached the first time?
<__monty__>
No, I mean secure firmware updates aren't a stretch goal.
<__monty__>
People might consider that more important than on-device TOTP.
<bbigras>
ah ok. because I thought a good stretch goal could be something that they need money to pay someone to do, but they could just do it way later in an update if the stretch goal isn't met.