vika_nezrimaya has quit [Ping timeout: 265 seconds]
<gchristensen>
*sigh* my gpg key has expired and I can barely motivate myself to renew it
<ivan>
"what's a gpg key" child voice
<ivan>
I threw mine away a few years ago
waleee-cl has quit [Quit: Connection closed for inactivity]
<qyliss>
Important for your role on the security team, no?
<gchristensen>
bingo
* colemickens
imagines (jokingly) GPG-as-a-service. Login with Facebook, we manage the key for you.
<gchristensen>
oof
<worldofpeace>
qyliss: Waah, just noticed that too. No more cool links for me to stalk other distro's
<gchristensen>
qtbase is such a big crompile
<worldofpeace>
C+++ and cmake + a very big project
<samueldr>
crompile somehow sounds more right than compile for qtbase
<worldofpeace>
precisely samueldr, I thought nothing of the phrase. accuracy 🏹
rardiol has quit [Ping timeout: 252 seconds]
drakonis has quit [Quit: WeeChat 2.6]
das_j has quit [Remote host closed the connection]
das_j has joined #nixos-chat
soju has joined #nixos-chat
buckley310 has joined #nixos-chat
soju has quit [Quit: Connection closed for inactivity]
ivan has left #nixos-chat [#nixos-chat]
<sphalerite>
pie_: you can also specify script = "foo"; instead of serviceConfig.ExecStart = "${pkgs.writeScriptBin "name" "foo"}/bin/name";
pie_ has quit [Ping timeout: 268 seconds]
vika_nezrimaya has joined #nixos-chat
veske has joined #nixos-chat
__monty__ has joined #nixos-chat
__Sander__ has joined #nixos-chat
<__monty__>
Is there a way to close a process' stdin without gdb?
vika_nezrimaya has quit [Read error: Connection reset by peer]
rardiol has joined #nixos-chat
<infinisil>
__monty__: I'm thinking something in /proc/$pid
sphalerite has quit [Ping timeout: 246 seconds]
sphalerite has joined #nixos-chat
pie_ has joined #nixos-chat
drakonis has joined #nixos-chat
pie_ has quit [Ping timeout: 258 seconds]
pie_ has joined #nixos-chat
tilpner has quit [Quit: tilpner]
veske has quit [Quit: This computer has gone to sleep]
<sphalerite>
__monty__: to what end?
<pie_>
sphalerite: @ script = foo, thanks
<__monty__>
sphalerite: Debugging an interactive program that supposedly doesn't quit when its stdin closes.
<__monty__>
gdb on macOS requires too much code signing nonsense for me to deal with for such a simple thing.
<sphalerite>
oh lol
<sphalerite>
maybe lldb?
<__monty__>
Same code signing things.
<sphalerite>
but doesn't that ship with xcode?
atlas_ is now known as shatlas
shatlas is now known as atlas_
pie_ has quit [Ping timeout: 268 seconds]
waleee-cl has joined #nixos-chat
pie_ has joined #nixos-chat
<pie_>
sphalerite: more networking race condition problems: i think i have the following situation: the container is not up until tinc is up, i want tinc to not be started until networking is up because i need to fetch some configs from the network, but for some reason the network doesnt come up untul tinc is up
<pie_>
* sphalerite: more networking race condition problems: i think i have the following situation: the container is not up until tinc is up, i want tinc to not be started until networking is up because i need to fetch some configs from the network, but for some reason the eth0 doesnt come up until tinc is up
<pie_>
the containers network stuff seems to be done by networkd, the config of which has an entry for the vpn device but nothing for eth0
<pie_>
sphalerite: i still dont quite know how my irc client works if i havent fully connected yet, did you get my messages
<sphalerite>
pie_: yep they all went through
<pie_>
ah ok
<pie_>
basically im not sure how to break the dependency of eth0 coming up on the vpn starting the device (attempting and failing to connect)
<sphalerite>
well, all the ones that went through went through lol
<pie_>
at least, i think thats what going on
<pie_>
i now have a file full of a page of commented out garbage code x'D
<sphalerite>
> eth0 doesnt come up until tinc is up
<{^_^}>
undefined variable 'eth0' at (string):271:1
<sphalerite>
how?
<sphalerite>
and how are you determinin gthat?
<pie_>
sphalerite: if i give tinc after = [ "systemd-networkd-wait-online.service" ] then https://bpaste.net/show/3DRKM
<pie_>
and i cant ping the internet from the container, it looks like the virtual device outside the container doesnt exist yet
<pie_>
i think maybe it only comes into existence after the container is up? that seems screwy though
<sphalerite>
pie_: you may need to ensure that tinc.retiolum isn't "required for online"
<sphalerite>
otherwise it'll be waiting for itself
<sphalerite>
I suspect that's what's happening there
<pie_>
right
<pie_>
yeah i suspect something like that as well
<pie_>
let me check the tinc service source
<sphalerite>
but it's better if you don't have it wait for networkd to say it's online, instead trying to access whatever resource it needs until that succeeds
<pie_>
sphalerite: i think thats kind of crappy but thats also what i tried originally and it still breaks it because: it blocks until its up
<pie_>
while [ "$(ping -w 30 -c 1 somehost.us > /dev/null; echo $?)" != 0 ]; do echo hi; done;
<pie_>
alternatively, and this is what i was originally trying to fix:
<pie_>
if i dont do anything, everything works except i have to restart tinc once for it to actually work
<pie_>
idk why
<sphalerite>
probably because that needs DNS?
<pie_>
yeah but thats the point, the bash script is trying to wait for network access
<sphalerite>
also, you can write the loop condition as `while ! ping -w30 -c1 somehost.us; do …` :)
<pie_>
ok
<sphalerite>
so why do you want to start tinc only after you have internet access again?
<pie_>
it needs to download some config files
<pie_>
im _assuming_ thats why i have to restart it once, its weird tho
<pie_>
well not tinc but the launcher script for tinc in this case
<sphalerite>
ah. Would it maybe work better if you had the launcher script separate from the tinc daemon itself, and make it ask tinc to reload its configuration once it's done?
<sphalerite>
that is, have an extra unit called download-retiolum-config or something
<sphalerite>
you can then either put `/run/current-system/sw/bin/tinc.retiolum reload` at the end of that (probably better, since then the tinc interface will be available right away) or give tinc an After dependency on it
<pie_>
sphalerite: probably, but that still seems like a crappy solution :(
<pie_>
i mean, if thats what it takes to work...
<sphalerite>
what's crappy about it?
<pie_>
reloading the service seems like a hack
<sphalerite>
I don't see how. The configuration's been changed, so we reload it.