<ajs124>
slighly related: the postgres config I copied for hydra recently seems to have worked! we're seeing less weird issues with, from what I can tell, which is nice.
<gchristensen>
nice
<ajs124>
so thanks again for bearing with me, while I was trying to figure out how the nixos-org config works ^^
<gchristensen>
:) I'm glad you sorted it
<ajs124>
It's hydra, though. There's always some new and exciting issue around the corner :P
<gchristensen>
:)
<ajs124>
and if hydra is actually working, you run into stuff like #84352 or #83782
parsley936 has quit [Remote host closed the connection]
<aanderse>
well dang
<aanderse>
steam didn't cloud save lego dc villians :S
<aanderse>
my kids aren't going to be happy if they find out we have to play all the levels we already beat again
<ajs124>
can't you download a savefile?
<aanderse>
steam didn't upload it!
<aanderse>
i'll have to dig through old hard drive and see if i can find it there
<cole-h>
(I think they meant download an already-complete one off the interwebs)
<aanderse>
that doesn't sound fun
<cole-h>
aanderse: Maybe could see if there's a save editor?
<emily>
00:04 * colemickens keeps catching himself with two different matrix clients up side-by-side, actively going back and forth :/
<emily>
colemickens: I do this because weechat-matrix is terrible at files and formatting and can't do multi-line messages :')
<colemickens>
latest commit is:"treewide get rid of go 1.12". I'm sad it's not "Let Go of Go"? "Let Go Go"? "Let Go, Go"?
<cole-h>
"Let It Go (1.12)"
* colemickens
just realized he isn't getting any notifications for colemickens. Time to revert this name change.
<MichaelRaskin>
«Let it Go, (but _not_ 1.12)»
<colemickens>
cole-h I think thats the joke I was aiming for and missed, haha
<cole-h>
aanderse: Did a quick Goog, and doesn't appear to have a save editor :(
<cole-h>
Sounds like it's time to stay up late and redo it... Think of the children!
<cole-h>
colemickens: Sorry to steal your thunder :P
<colemickens>
not at all
drakonis has quit [Quit: WeeChat 2.8]
<cole-h>
colemickens: 👀 Looks like I just got my first organic star on passrs... A Red Hatter, no less
<colemickens>
oh, that's a former coworker. I wonder if he follows me on gh
<colemickens>
not that he couldn't have found it on his own, sorry that was rude
<cole-h>
:P
<cole-h>
I didn't take it that way
<cole-h>
Still, it was organic in that I didn't advertise it!
<samueldr>
cole[mickens] advertising for cole[-h] seems like cole-usion to me
<cole-h>
>:D
<cole-h>
samueldr++ That was puntastic
<{^_^}>
samueldr's karma got increased to 221
<cole-h>
btw colemickens, before you PR the libsecret stuff (whenever that happens; no rush), it would be great if you could find a way to use dbus-codegen in build.rs, rather than having to install the CLI tool
<ashkitten>
i wonder what percent of all karma in this channel is for puns and what percent is for helpfulness
<gchristensen>
hehe
<cole-h>
Well, 99% of my ++ in here is definitely for samueldr's puns
<MichaelRaskin>
We need full /. with separate Insightful and Helpful and Funny karma.
<ajs124>
Whatever happened to /.? Are they still around? Does anyone still read that?
<ajs124>
Huh, they still are. Their design looks… basically, like it always did, except newer.
drakonis has joined #nixos-chat
<colemickens>
I feel like ofborg might want to not notify reviewers for drafts?
<gchristensen>
interesting idea
<gchristensen>
nice, found a reproducible segfault in bash
<cole-h>
That's not good
<samueldr>
quick, find a name, register the domain name and profit
<cole-h>
lol
<gchristensen>
it has dumped core since 2014... they probably know about it
<gchristensen>
bash -c "trap 'kill -- \$\$' INT TERM QUIT; set -x; kill \$\$; sleep 50"
<cole-h>
6 years???
das_j has quit [Quit: killed]
ajs124 has quit [Quit: killed]
ajs124 has joined #nixos-chat
das_j has joined #nixos-chat
<elvishjerricco>
What in the world does this bash syntax do? `set -- $(IFS==; echo $o)`
<gchristensen>
[grahamc@Petunia:~]$ export o="foo=bar"; echo set -- $(IFS==; echo $o)
<gchristensen>
set -- foo bar
<ornxka>
huh
<ornxka>
so bash does $IFS even when you just dereference a variable
<gchristensen>
it is unquoted when passed to echo, so "evaluated"
<gchristensen>
oh elvishjerricco I see
<gchristensen>
[grahamc@Petunia:~]$ set -- foo bar
<gchristensen>
[grahamc@Petunia:~]$ echo "$@"
<gchristensen>
foo bar
<elvishjerricco>
Wait what on earth is $IFS?
<gchristensen>
Internal Field Separator
<ornxka>
special bash delimiter variable
<ornxka>
whoa so thats what it stands for
<elvishjerricco>
How does it work?
<gchristensen>
IFS defines what characters cause word splitting
<ornxka>
IFS=$'\n'; for foo in "bar bar" "qux foo"; do echo $foo; done
<elvishjerricco>
Oh. So `IFS==` sets the delimiter to `=`. Then `echo $o` splits `foo=bar` by said delimiter, thus passing `foo` and `bar` as separate arguments?
<infinisil>
Ew
<gchristensen>
yeah
<elvishjerricco>
Yea, jank, but useful
<ornxka>
without assigning IFS it will print bar\nbar\nqux\nfoo; with assigning IFS it will print bar bar\nqux foo
<gchristensen>
you shouldn't use it, as demonstrated by the last 7 minutes :P
<elvishjerricco>
Lol
<elvishjerricco>
It's the argument parser in at least stage-1.sh
<infinisil>
I'd probably use `cut -d= <<< "$o"`
<gchristensen>
oh yeah ther eyou go
<elvishjerricco>
Gettin' back into the systemd initrd stuff
<elvishjerricco>
Gonna see if I can rip out all the device/mounting stuff and turn them into systemd-isms
<infinisil>
Oh wait cut doesn't work
<infinisil>
Is there a nicer way to make this in pure bash?
<gchristensen>
IMO, not really
<gchristensen>
my answer would be "don't use bash for this" but stage-1 makes it harder
<infinisil>
Hm, is there something that could be used that low that's not as awful as bash?
<elvishjerricco>
I remember stdenv used to depend on a GCC argument parser written in bash... That thing was so slow one of our projects took like 15 minutes longer to build
<elvishjerricco>
don't write parsers in bash :P
<gchristensen>
yeah but also how often do you pass thousands of arguments to your kernel lol
<elvishjerricco>
The funny thing when debugging it was trying to strace it and just seeing endless `brk` instructions
<infinisil>
It would be really cool if nix-copy-closure (or some wrapper around it) could intelligently decide where to get paths from based on interface speed
<gchristensen>
or used all of them!
<infinisil>
I mean, currently you can pass --use-substitutes, which will make it fetch things from cache.nixos.org if the paths are present in it
<infinisil>
I guess I don't mean interface speed
<infinisil>
But host connection speed
<gchristensen>
yeah but what if it did both
<infinisil>
Yes that's what I'm thinking too :)
<infinisil>
But it would have to decide which one to use for which paths to make it as fast as possible
<infinisil>
So it's not entirely trivial (but I'd expect it to be pretty simple still)
<infinisil>
Oh, imagine this: It starts downloading from both cache.nixos.org and localhost like crazy, but then cache.nixos.org goes down, which it notices because speed goes to 0, at which point it switches to using localhost for everything
<infinisil>
That would be sweet
<infinisil>
I mainly want this because currently it's annoying to have to decide whether to use --use-substitutes or not
* infinisil
is currently waiting on a nix-copy-closure..
<infinisil>
It's kind of amazing how nix#8 doesn't have any reasonable solution yet
<emily>
03:37 <elvishjerricco> Gettin' back into the systemd initrd stuff
<emily>
elvishjerricco: oh nice! i was considering working on a greenfield initrd implementation based on systemd
<emily>
that works a lot more like just "a normal nixos system"
<emily>
interested in seeing whatever you come up with
<emily>
(sadly nixos modules' complete lack of support for instancing makes this a lot more code and work than it ideally should be)
<elvishjerricco>
emily: I did it that way once. Literally eval'd a special nix config and pushed it into a cpio.
<elvishjerricco>
It was like 60M lol
<elvishjerricco>
AFTER manually removing a bunch of stuff
<emily>
what if you used musl?
<elvishjerricco>
emily: I've already got a PR that's more iterative on the current system that has /init = systemd in initrd
<elvishjerricco>
emily: Can't. Systemd depends on glibc really hard
<elvishjerricco>
Plus glibc wasn't really a huge problem. It was more that shitloads of unnecessary packages were there because NixOS does a lot by default
<emily>
you could use musl for everything but systemd, but yeah
<emily>
mhm
<emily>
I mean, you'd really not want to use "standard" nixos if that pulled in normal coreutils and the like
<elvishjerricco>
Wouldn't be much point in replacing glibc with musl if you're just gonna add glibc anyway :P
<elvishjerricco>
I'm actually considering replacing busybox with coreutils and util-linux. Should only be one or two megabytes bigger that way
<elvishjerricco>
Would make a lot of things easier
<emily>
fair enough -- a more similar environment in initrd would be nice
<emily>
especially for debugging boot
<emily>
tbh, nixos initrds are already huge so I wouldn't worry too much
waleee-cl has quit [Quit: Connection closed for inactivity]
<emily>
elvishjerricco: do you know if systemd supports "transferring" the supervision tree when executing the stage-2 systemd in some way?
<emily>
elvishjerricco: thinking of, e.g. keeping a gpg-agent alive from initrd into the actual system
<emily>
it'd be nice to do that as a normal systemd service with no gross hacks
<elvishjerricco>
emily: Yea it does exactly that if stage 2 init is exactly the path to systemd
<elvishjerricco>
It bakes its installed path into the binary when it's built and checks that
<emily>
nice, so you can just install the gpg-agent service into initrd, and it'll persist beyond stage-2 (as long as you make sure it's not referencing stuff on the dead fs etc. ofc)?
<elvishjerricco>
That's how plymouth starts so much earlier in other distros
<emily>
maybe you could even have a generic option for it? systemd.services.foo.initrd = true;
<elvishjerricco>
I'm not sure on the specifics. Check out the man page for systemctl in the switch-root section
<emily>
that might let you reuse more existing code in the initrd as well
<elvishjerricco>
I do want to have a NixOS option similar to systemd.packages that copies the unit files in and checks them for binaries to also copy
<elvishjerricco>
Then the plymouth module would be dead simple
<emily>
btw, you should take this as an opportunity to rename boot.initrd to boot.initramfs, because it hasn't been an initrd for a long time >_>
drakonis has quit [Quit: Leaving]
<colemickens>
emily: definitely curious what you're thinking about doing with the gpg-agent in initrd and/or it's important it survive the stage1->2 transition
<emily>
colemickens: enter yubikey pin at startup, have it automatically decrypt zfs encryption key, etc., and persist into my user session
<emily>
just the eternal quest to reduce the number of passphrases I need to memorize to 1 :)
<emily>
I guess it might stay unlocked even between gpg-agent invocations?
<cole-h>
If you ever work this out, be sure to put out a blog post or post to the Discourse... I'd be interested to do the same thing (after getting my key, that is)
<emily>
will try! making emily.moe support actual content is, uh, at some random position in the adhd todo heap and will be swapped in in approximately time period
<cole-h>
Hehe
cole-h has quit [Quit: Goodbye]
pie_[bnc] has quit [Ping timeout: 260 seconds]
pie_[bnc] has joined #nixos-chat
<sphalerite>
armin: another key feature of guix is that it has much, much better UI than Nix
tilpner has quit [Quit: tilpner]
parsley936 has joined #nixos-chat
Synthetica has joined #nixos-chat
<hyperfekt>
sphalerite: are you talking about the CLI? or what part?
avn has quit [Ping timeout: 246 seconds]
avn has joined #nixos-chat
<armin>
sphalerite: hmm, i have i3 and kde on nix, which are both pretty good UIs.
vika_nezrimaya has joined #nixos-chat
<armin>
sphalerite: i mean no offense, but that's very un-specific, i don't have any problems UI-wise with nixos.
__monty__ has joined #nixos-chat
vika_nezrimaya has quit [Ping timeout: 256 seconds]
<sphalerite>
hyperfekt: yes
<sphalerite>
armin: the guix CLI is much better than the nix CLI
<sphalerite>
it feels like it's thought out as opposed to organically grown
<armin>
sphalerite: ok, but what exactly is guix? is it like a fork of nixos? is it just nixos meant for desktops?
<armin>
sphalerite: should i even bother with it when i'm happy with nixos?
<sphalerite>
guix is a package manager built on the nix model and using the nix daemon, but with its own frontend — no nix language, it uses scheme instead
<sphalerite>
guixsd is the nixpkgs and nixos equivalent of the guix world, written entirely in scheme (no bash involved, which is also nice)
<sphalerite>
(this is my understanding, I haven't actually used guix much)
<armin>
sphalerite: well i'm a hardcore bash scripter but i understand why you wouldn't want to have bash in your package management tools. :)
<armin>
sphalerite: thanks for your kind replies, though.
<sphalerite>
I'm happy with nixos too, but especially when helping other people with nix its UI problems it's quite clear that nix's UI is a bit awful
<sphalerite>
of course, there's work in progress to improve that, but guix is just miles ahead in that respect
<armin>
ok sounds reasonable, thanks again.
<sphalerite>
it also seems to have better organised documentation from what I've seen
<armin>
ah yeah that nix manual that's gambling around in pdf form of the whole web page on my computer since weeks.
<armin>
and sometimes i stumble upon these aspects of nix where i'm like "ok i don't understand a word of what's written there".
<armin>
that being said, i was able to almost fully automate the deployment of my laptop computer with full disk encryption and whatnot, it's working splendid.
<sphalerite>
one major practical issue with guix is the hardline stance on proprietary software, which can make it more difficult to use various hardware
<armin>
like spotify for example?
<MichaelRaskin>
Who cares about spotify
<MichaelRaskin>
Like Intel WiFi firmware
<armin>
i do, i record spotify to cassette tapes even.
<armin>
consider me mad.
<MichaelRaskin>
I mean, you can always do some silly trick for installing Spotify in a Guix-managed FHS-like chroot or something
<MichaelRaskin>
With WiFi firmware, reasonable integration with the system is more desirable
<eyJhb>
Yay or nay, in a big company dealing with sensitive information at times, about 24.000 employees
<cole-h>
Looks like I need to switch to PGTK Emacs again... XWayland Emacs keeps on hanging my computer (:
<joepie91>
eyJhb: very nay
<joepie91>
eyJhb: Zoom is *absolutely not* designed to be private or secure by design, whatsoever
<clever>
ive heard of people brute-forcing zoom meeting id's, and then bombing random classes for school
<eyJhb>
What if hum really wants it joepie91 ? And can't use Teams?
<cole-h>
lovesegfault: lol, I switched from waybar to swaybar temporarily because of the gtk issue... now that I've rebuilt gtk, I think I'm gonna stick with swaybar... :P
<joepie91>
eyJhb: cannot parse that sentence
<joepie91>
"what if hum really wants it"
<joepie91>
syntax error at line 1, position 10
<cole-h>
h..... upper management?
<joepie91>
er, 8, sorry
<eyJhb>
Hum => department of human resources etc. :p
<joepie91>
eyJhb: I mean, really wanting something doesn't change whether it's secure or private?
<lovesegfault>
cole-h: :D
<joepie91>
so the answer doesn't change :P
<cole-h>
lovesegfault: I tried writing a minimal MPD application in C that gave me what I wanted, but it ate too many resources... :( And async C is witchcraft, so I didn't even try MPD's async bindings
<eyJhb>
I suggested that, they did not care. Thought it was a discussion of opemsource vs. buyware, as it was called
<joepie91>
eyJhb: might want to ping Legal
<joepie91>
and tell them that using Zoom constitutes legal risk to the company due to its poor handling of potentially-confidential information
<colemickens>
eyJhb: I don't envy the position you're in.
<joepie91>
and let them pick the fight :P
<eyJhb>
The data agreement is in place, for GDPR. And the servers used is not hosted by Zoom, but a provider
buckley310 has joined #nixos-chat
<colemickens>
joepie91: the problem now is that they just put out an update, so this fight just became even more impossible
<eyJhb>
But that does not make the software more secure
<colemickens>
joepie91: because now every criticism is "oh well they fixed it"
<joepie91>
eyJhb: data agreement isn't enough :P
<colemickens>
never mind the incentives, motivations, priorities, unchanged fundamental nature, etc, blah blah, it still doesn't matter
<joepie91>
colemickens: yeah that's how this shit usually goes, because ~noone understands that security is a structural thing and not a matter of fixing specific issues
<joepie91>
yeah that
<eyJhb>
If anyone wants to send me the links for all the security things the past 6-12 months, I would appreciate it.. Have to make dinner. But there is like 5? But yeah. The position sucks. It is arrogant of them. And there is sooo much
* colemickens
glances over at the landing page for a dev community last night that he is just passing on because they only exist on slack
<qyliss>
lol, I'd hope all the security things for the past month would be enough
<MichaelRaskin>
I am nostalgic of times when the mere «macOS installer» story about Zoom was sufficient to declare them intentional malware
<eyJhb>
It is STUPID. I really do NOT get the logic
<savanni>
And yet, I have no idea what I would move my company, community meetings, or interviews to without causing huge disruption. Interoperability is a big deal.
<ldlework>
What's the command I need to prefix with to use a gpu when an integrated chip is present?
<ldlework>
I thought it was "bumblebee" but apparently not.
<lovesegfault>
cole-h: not yet, I think I can salvage it
<qyliss>
actually, just read that schneier post
<qyliss>
it seems to cover a bunch of them
<qyliss>
you can extract all the links from there to send to people, probably
<cole-h>
lovesegfault: Without that stdenv lchmod whatever patch? :P
<savanni>
cole-h: Like I said to gchristensen a few days ago... I play on OGS, too. DDK ranking, though.
<cole-h>
:o More Go friends!
buckley310 has joined #nixos-chat
<lovesegfault>
cole-h: No, but I argue that the correct thing to do is to add ac_cv_func_lchmod=no to stdenv
<cole-h>
You'll need to convince fl*kli and co of that, then :P
<lovesegfault>
Yeah, we'll chat on mumble soon
<LnL>
unless we know what's going on I also feel like reverting would be better
<lovesegfault>
LnL: I _think_ I know what's going on
hax404 has quit [Ping timeout: 240 seconds]
hax404 has joined #nixos-chat
<lovesegfault>
I don't know why binutils is causing it, but I think I know _what_ it's causing
<lovesegfault>
glibc implemented their own lchmod, but it requires /proc to be mounted for it to work. The binutils bump caused autotools to detect lchmod as being available, but /proc isn't mounted in the build sandbox, so when things get built as if there was lchmod support their tests fail b/c they can't run the lchmod inside the sandbox
<lovesegfault>
this, for example, breaks `cp` of coreutils
<eyJhb>
qyliss: it is useless, there is not much to do I guess.
<lovesegfault>
The thing is that Linux _does not_ have lchmod, and glibc's half-implementation does more harm than good, so I'm arguing we should default autotools to _not_ use lchmod on Linux by adding `ac_cv_func_lchmod=no` in the stdenv
<MichaelRaskin>
GH UI update vs consistency: previously, merging a PR changed the status icon on the top of the PR, now this thing seems to fail
<gchristensen>
"
<gchristensen>
The difference between fine and great software is listening to people.
<gchristensen>
oops
<cole-h>
"
<ornxka>
yesterday i spent several hours trying to figure out why encryption wasnt working on lineageos because instead of providing helpful log messages during boot it just displays the los logo
<ornxka>
i think this means the difference between fine software and awful software is just the absence of willful malice
<MichaelRaskin>
Unfortunately, there are people who actively do not want to be shown information even when they are free to ignore it
<MichaelRaskin>
It looks like a lot of horrible software messes of recent are built by cherry-picking the set of users to listen to.
<srk>
hmm, does anyone know how to overcome problem with kernel patches unable to create new files?
<eyJhb>
Anyone good at probability math (pmf right now)?
<MichaelRaskin>
What exactly do you want?
<eyJhb>
Unsure how to find the PMF of X, if I have a given propability of something happening
<eyJhb>
Can send the desc
<eyJhb>
A message is transmitted from A to B. Probability to receive a message is 0.8. A message is retransmitted until it is successfully received. Let X be a random variable represented a number of time it took to send the message until it is successfully received.
<MichaelRaskin>
You mean you have a condition?
<eyJhb>
Find pmf of X.
<eyJhb>
Y
<eyJhb>
MichaelRaskin: does that make better sense?
<MichaelRaskin>
Ah
<MichaelRaskin>
Well, can you find the probability of X=1? Next question: X=2? Next: X=k?
<MichaelRaskin>
(Wikipedia says that the answer to the third question is just by definition the same thing as PMF)
<eyJhb>
Third question? Find PMF of X?
<MichaelRaskin>
k → Pr[X=k] seems to be what they call PMF
<MichaelRaskin>
So the question is which of the questions: Pr[X=1], Pr[X=2], Pr[X=3], Pr[X=k] is the first one where you have difficulties
<genevino>
so since this is getting a bit chatty, i'm asking in -chat
<genevino>
what EXACTLY is happening there? i know this makes a new environment in which the packages specified with -p exist, but i want to know more
<ashkitten>
infinisil: it'd be cool if i could see the output from the activation script when deploying with nixoses
<infinisil>
Yeah that would be nice, will definitely implement this eventually
<infinisil>
Currently it's a pain to see what went wrong
<elvishjerricco>
Upside of spending 6 hours upgrading my desktop to 20.03: The work transferred over to my other machines, and they worked on the first try.
<cole-h>
How many other machines?
<elvishjerricco>
cole-h: Just two.
<cole-h>
Then really, you only spent 2 hours upgrading :D
<elvishjerricco>
:P
<elvishjerricco>
On the other hand, my MacBook's graphics driver just spontaneously crashed and panicked my whole system. So that's cool
<cole-h>
Welp.
<armin>
welllllllllllllllllllllllllllllp
<samueldr>
but I also heard elvishjerricco makes it harder on themselves on purpose ;)
<samueldr>
(like an involved setup with conditional gpu passthrough)
<samueldr>
so I would say that 6 hours for a major system upgrade under those conditions is commendable!
<elvishjerricco>
samueldr: Actually the hardest part was figuring out why my stupid theme setup didn't work anymore :P
<elvishjerricco>
GPU passthrough worked no issue!
<samueldr>
hah
<samueldr>
heh the trivial things that one can't live without :)
<elvishjerricco>
But the mac is fairly close to vanilla; just nix and tinc are out of the ordinary. So that's apple's fault
<cole-h>
elvishjerricco: 👀 You use passthrough?
<elvishjerricco>
unless the custom kext that tinc depends on caused it... oh god
<elvishjerricco>
cole-h: "use" is a strong word lol
<cole-h>
But you have it working
<elvishjerricco>
I set it up because I thought it'd be cool and I thought I'd use a Windows VM with it
<elvishjerricco>
But then proton came out like the next week
<elvishjerricco>
and I basically haven't used it at all
<elvishjerricco>
But yea it works
<elvishjerricco>
I can even kexec-reboot into it
<cole-h>
Heh. Unfortunately, most of my games have anticheats, so Proton is a no-go.
<cole-h>
And also I use nouveau so I can run sway :P
<elvishjerricco>
I only have one GPU, so my passthrough setup is a `nesting.clone` system, where the clone boots up a windows vm with the gpu passed through. Kind of a pain to get linux not to take control of the main gpu, but it works. So this system is more like a dual boot, except I didn't want to actually put windows on my real hardware because gross :P
<cole-h>
I also only have 1 GPU :o
<cole-h>
I use qemu hooks to {un,re}hook linux from graphical stuff
<elvishjerricco>
Oh there was one quirk with the update
<elvishjerricco>
The mouse no longer works in the vm :P
<elvishjerricco>
Not sure why. Probably need to change some id or something
<elvishjerricco>
I bet I know what it is... I plugged in new usb devices. I bet they rearranged some ids or something like that.
<cole-h>
Heh, I send in my entire USB bus
<elvishjerricco>
"I use qemu hooks to {un,re}hook linux from graphical stuff" what do you mean by this?
<elvishjerricco>
cole-h: I did that at first, but then I realized that showed windows my external backup disk, and I did not like that :P
infinisil has quit [Quit: Configuring ZNC, sorry for the joins/quits!]
<cole-h>
elvishjerricco: One sec, pulling up the scripts
<samueldr>
I believe one of those should be a 10 on the CVE scale
<gchristensen>
wow
<gchristensen>
wow!
<samueldr>
at least last week it was reported as an upcoming 10
<gchristensen>
WOW.
<cole-h>
WOW!
<samueldr>
remember before shitting^W condescendingly write about anything, that's not well liked here
<cole-h>
I'm just saying 'WOW!', those sound pretty serious 👀
<gchristensen>
definitely
<gchristensen>
big mistakes like this can happen anywhere, for sure
<samueldr>
oh, that definitely wasn't about what you said :)
<samueldr>
though here it doesn't happen often, it is apparently the internet pass-time of excellence to talk down anything you don't personally use and agree with
<samueldr>
(without even needing to understand it!)
<cole-h>
samueldr: I don't understand you, therefore I hate you.
<cole-h>
(jk <3 samueldr )
<{^_^}>
samueldr's karma got increased to 222
<__monty__>
Yikes.
<samueldr>
I don't agree with you cole-h, I like myself you can't hate me, therefore you must be wrong
<samueldr>
(jk)
<cole-h>
:D
<__monty__>
Didn't they drop a custom transport because it was too hard to do right? Is there no off the shelf authentication that would've worked?
<genevino>
so does anyone of you build customized ISOs for nixos?
<samueldr>
I have no idea, I know next to nothing about salt stack's details, I only was reporting on it since there are sysadmin-like people here
<emily>
samueldr: fwiw the things I've heard from people who understand and work with salt don't reassure me any more :P
<cole-h>
ldlework: 👀
<__monty__>
I thought Salt was Puppet, but better.
<emily>
(but also, thanks, passed on to such a person)
<cole-h>
Also, since we have it packaged, maybe there should be one of those fancy security issues filed
<samueldr>
__monty__: that CVE doesn't mean your preconception is wrong ;)
<__monty__>
samueldr: Makes it less clear-cut at least.
<genevino>
adisbladis: just curious.
<cole-h>
I swear, the thing that's gonna make me switch to NixOS first is the fact that Nix-built Emacs on non-NixOS segfaults for me in my notetaking org mode buffers >:(
<samueldr>
>> There is a more advanced version of Popcorn Linux that allows applications to concurrently execute across ISA-different cores (.e.g. X86, ARM).
<samueldr>
!!
<cole-h>
The interesting thing is that it's reproducible on this one tree, for me...
<samueldr>
imagine being able to move the firefox window from your desktop to your phone...
<samueldr>
... though heh, imagine all the resources you need to track
<MichaelRaskin>
Today I imagined a review process where reviewer is not allowed to check out code locally. If there is something to test, well, there should be a test and something like ofborg will run it.
<MichaelRaskin>
(Consider this a drunk musing, except I do not consume external ethanol in any interesting concentrations)
<pie_[bnc]>
certified CI
<pie_[bnc]>
paper CI on the blockchain
<pie_[bnc]>
(im sorry)
<pie_[bnc]>
- half serious
<MichaelRaskin>
What exactly the blockchain would be about here …
<pie_[bnc]>
or just secure timestamping, but not really needed
soju has joined #nixos-chat
<MichaelRaskin>
Arghm
<MichaelRaskin>
I imagined a blockchain version of it
<MichaelRaskin>
Where each node, on receiving a block, rebuilds and reruns all the corresponding tests.
<emily>
cole-h: join the dark side, what are you waiting for? :3
<cole-h>
I think I just tracked down my issue, thanks to that backtrace I posted earlier... My Iosevka Custom seems to have been doing something strange -- changing to DejaVu Sans Mono and no problems where it was previously reproducible
parsley936 has quit [Remote host closed the connection]
<gchristensen>
adisbladis: d'you have any links for me, w.r.t. nice emacs things to do before tomorrow?
<infinisil>
I recently thought of a potentially stupid but interesting idea: The karma you give has a worth equivalent to the karma you have
<MichaelRaskin>
Linearly?
<infinisil>
Oh no, I meant s/equivalent/proportional
<MichaelRaskin>
At least with +1?
<cole-h>
I was gonna say, what about the newbies who are thanking me for my help? :(
<MichaelRaskin>
cole-h++
<{^_^}>
cole-h's karma got increased to 34
<gchristensen>
newbies giving karma seems extra valuable
<cole-h>
Exactly, cuz nobody knows about it
<infinisil>
Hm yeah, so maybe it really is a stupid idea :P
<cole-h>
Which is why I'm so cavalier with my karma-giving :^)
<cole-h>
I want them to know it exists, so they can give me more of it >:D
<gchristensen>
on the other hand, my amount of karma feels like too much relative to so many people whose karmas are not so high
<gchristensen>
(I've sort of wondered about making it ambigious ("xxx's karma increases")
<gchristensen>
...)
<infinisil>
logarithmic karma
<infinisil>
"foo's karma still has 1 digit"
<MichaelRaskin>
I considered parsing «thank you» for karma, but then grepped the logs…
<MichaelRaskin>
infinisil: first of all, it deincentivises talking to newcomers.
<adisbladis>
Considering the size of emacs packages I was surprised we don't have more failures
<MichaelRaskin>
Because people who are new-ish here are more likely just not to know that «nickname <3» and «<3 nickname» are different messages etc.
<MichaelRaskin>
This is the bad part.
<gchristensen>
gotta keep things happy and fun, being scored on usefulness isn't so fun
<cole-h>
^ Especially that
<adisbladis>
Karma is useless anyway :P Let's not pretend otherwise
<infinisil>
Hehe yeah, even the current karma can be a bit annoying
<cole-h>
I always get a kick out of seeing someone do "✨ worldofpeace" (or insert any other name), tbh
<gchristensen>
adisbladis: definitely
<cole-h>
Oh no I forgot to censor that one
<MichaelRaskin>
The _ugly_ part is that answering a substantial question from an experienced person here likely gives you _one_ +1, and a laser-guided pun might give you 2 or 3
<cole-h>
Well...
<gchristensen>
adisbladis: for example, aszlig++ isn't really *so* terrible.
<{^_^}>
aszlig's karma got increased to -666
<infinisil>
MichaelRaskin: True :)
<cole-h>
idk just seeing that number... aszlig doesn't seem like a very trustworthy person.... :P
<MichaelRaskin>
gchristensen: re: fun — well, if you make it absurd enough, it is still fun! But it is better not to create completely crazy incentives even in jest
<aszlig>
cole-h: i completely agree :-)
<cole-h>
aszlig: Hehehe.
<adisbladis>
"aszlig isn't really *so* terrible"
<adisbladis>
I'm inclined to agree
<infinisil>
Oh, maybe I should make the bot emit messages like "foo's karma got increased to like 200, can't remember exactly!"
<gchristensen>
what if karma followed the rules of numberwang
<adisbladis>
cole-h: It's funny how much that overlay cut down on my emacs maintainer time
<gchristensen>
beautiful
<adisbladis>
I've barely touched things in over a year
<adisbladis>
It maintains itself
<cole-h>
adisbladis: It's really good, thanks a lot for it. I like to run on the bleeding edge, as you'll notice when I complain about rebuilding the world every gtk/etc. update
<adisbladis>
=)
<adisbladis>
We have a binary cache now!
<cole-h>
gchristensen: Trying to play this game is making my brain hurt, send help
<adisbladis>
What are you complaining about ;)
<cole-h>
adisbladis: I mean more that I literally have to rebuild gtk when it fixes some issue
<cole-h>
And then I remember why it has red labels
<cole-h>
(e.g. recently, 3.24.17 -> 3.24.18 fixed an issue where waybar wouldn't show up, which just recently made it into the cache)
Synthetica has quit [Quit: Connection closed for inactivity]
<adisbladis>
Haha it seems that all my projects start with some tiny itch that snowballs
<MichaelRaskin>
cole-h: it is 2048 for illiterate, being able to read the numbers puts you at disadvantage
<cole-h>
I started building it myself, then realized how many tools I use that depend on gtk, and called it quits (and switched back to the default swaybar with stuff)
<adisbladis>
I just wanted up to date exwm ^_^
<cole-h>
MichaelRaskin: You're 100% correct lmao
<cole-h>
MichaelRaskin: Wait I figured it out
<cole-h>
It's all in the colors of the tiles
<cole-h>
:D
<MichaelRaskin>
Yes, that's the point
* cole-h
has only played 2048 a few times, many years ago
* infinisil
has become distracted with changing the karma implementation a bit
<cole-h>
I like how GH underlines the URLs but doesn't make them clickable 🙃
<adisbladis>
cole-h: I never really noticed
<adisbladis>
How can I unsee this
<cole-h>
🙃🙃
<cole-h>
^ that's us
<MichaelRaskin>
adisbladis: easy, switch to gitea
<aszlig>
or maybe just without a webui =)
<gchristensen>
yeah, adisbladis, flycheck is hitting Lisp nesting exceeds ‘max-lisp-eval-depth’ and I'm not too inclined on heading to the overlay :
<gchristensen>
:/
<MichaelRaskin>
infinisil: if you want to safeguard against karma becoming too serious business, maybe stupid tricks could help. Like if DM-ing «nickname++ #channel» to the bot worked and announced the change in the channel…
<aszlig>
ah, since we're nix(os) offtopic here: anyone happen to know a cross-platform way to find out the full path of getting the path of a linked shared library?
<joepie91>
cole-h: ha
* joepie91
adds a 🙃
<drakonis>
oho that's a name i havent seen in a while
<infinisil>
MichaelRaskin: Why would people `nickname++ #channel` to the bot?
<cole-h>
gchristensen: I `(setq max-lisp-eval-depth 400)` and I don't ever see that problem
<gchristensen>
lol cool
<cole-h>
Which is actually lower than the default, it appears... lol I guess that won't help
<MichaelRaskin>
infinisil: they will, after seeing the crazy abuse it allows!