<dtz[m]>
sorry for not putting into a PR right now
lassulus_ has joined #nixos-dev
lassulus_ is now known as Random_int
lassulus has quit [Ping timeout: 260 seconds]
Random_int is now known as lassulus
<aanderse>
aszlig: thanks for comments. happy to learn more.
<aszlig>
aanderse: thanks for your contribution
<aanderse>
i'm not super familiar with writing systemd units, but when you said "No need to use a shell here, ExecStartPost might be a better idea." did you mean ExecStartPre?
<aszlig>
hm, when seeing "with lib;" at the beginning... we have that and similar things accross a lot of nix expressions...
<aszlig>
aanderse: oh, whoops, yep
<aanderse>
ok, thanks
<aszlig>
so if checking the file with nix-instantiate --parse, something like with foo; for the whole file will not show errors for invalid identifiers
<aszlig>
if i would have a blog, i'd probably write an article about "with ...; considered harmful"
<aanderse>
guest post on nixos.org? :)
<aanderse>
hmm... changed to execstartpre and now get this: incron.service:12: Executable "mkdir" not found in path "/usr/local/sbin:/usr/local/
<aanderse>
/bin:/usr/sbin:/usr/bin"
<aszlig>
or maybe even: with "..."; considered harmful
<aszlig>
aanderse: ${pkgs.coreutils/bin/mkdir
<aszlig>
nix-instantiate --parse -E 'with "attr"; considered harmful'
<aanderse>
ah yeah... every single binary needs a path in there right
<aszlig>
aanderse: yep, but IIRC there was a recent systemd change so that relative paths are allowed as well
<aszlig>
but i'd recommend against that =)
<aanderse>
ok
<aszlig>
infinisil: does your bot have support for --parse?
<infinisil>
Nope
<aszlig>
infinisil: ah, but it is in EvalMode, so it could be added easily
<infinisil>
Indeed
<aszlig>
although i doubt it will be very widely used =)
<infinisil>
Have seen it occasionally where it could have been
<infinisil>
Oh and regarding `with ...;`, i reall don't like this:
<infinisil>
> let valll = "outer"; in with { valll = "inner"; }; valll
<{^_^}>
"outer"
<aanderse>
aszlig: ok everything you mentioned has been fixed up, except the forking
<aszlig>
aanderse: there is -f for running in foreground
<aszlig>
err
<aszlig>
-n
<aanderse>
aszlig: hmm yeah i could try type simple with the -f flag
<aszlig>
aanderse: yeah, in that case you can remove Type completely as "simple" is the default
<aszlig>
infinisil: yeah, that one is well known since quite a while... another reason ;-)
<aszlig>
infinisil: although there is a reason why it's implemented that way:
<aanderse>
erm -n flag i mean
<aanderse>
-n, --foreground
<aanderse>
would i still need the pid file?
<aszlig>
infinisil: if doing something like "with pkgs; [ a b c ]" and b comes from a let, you don't want to have a new package "b" suddenly override your binding
* aanderse
reads more about systemd
<aszlig>
aanderse: not necessarily
<aanderse>
ok, it builds, runs, and the tests i had still work the same
<aszlig>
aanderse: lemme check whether the pidfile might be used for startup status determination if type is simple
<aszlig>
aanderse: ah, you have NixOS tests?
<aanderse>
aszlig: no :( not yet
<aanderse>
my manual tests
<aanderse>
i started reading about tests the other day
<aanderse>
because the person who merged my redmine change asked for some tests
<aanderse>
still have a bit more work to do on that
<aanderse>
and reading
<aszlig>
aanderse: okay, the pidfile does set the state of the service to SERVICE_SUCCESS
<aanderse>
oh, ok
* aanderse
adds pidfile back
<aszlig>
aanderse: ah, wait...
<aanderse>
mhm?
<aszlig>
aanderse: okay, just wanted to check the point where the pidfile is created
<aanderse>
and aside from starting to bribe people with beer... is there anything i changes i can make or things i can do to possibly get this merged in so it will make it in for 18.09?
<aszlig>
aanderse: i went looking because some services do create the pidfile right at the start of main(), so it's kinda pointless
<aanderse>
ah i see
<clever>
man sd_notify
<clever>
that will probably solve everything, if your willing to rely on systemd
<aszlig>
aanderse: i'd wait for jtojnar to approve, then it's fine
<aszlig>
clever: that's not supported by incron
<aszlig>
clever: but if you're relying on systemd you don't need incron
<clever>
ah
<aanderse>
aszlig: awesome! thank you for all your review and help. i really appreciate it. can you possibly ping jtojnar and mention you're good to merge as long as they approve?
<aanderse>
jtojnar had mentioned they wanted someone else to give a definitive approval before they would approve
<aszlig>
clever: there's systemd.path(5) which can do the same
<aanderse>
ok, way past my bedtime. thanks again. good night.
<aszlig>
aanderse: well, for my part i'm a bit gruesome, because i usually don't merge something without a test
<clever>
aszlig: sd_notify seems more flexible, given that i can set a full status msg, and do watchdog stuff
<aszlig>
clever: more flexible than path units?
<clever>
in terms of stating when a service is started, stopping, and its status
<aszlig>
clever: yeah, that for sure
<aszlig>
clever: it's better than the pidfile approach, as you can also set status messages
<clever>
yep
<clever>
and you can have the process running, but not yet started
<aszlig>
but for services which do not support it the pidfile approach would be "okay"ish
<clever>
toxvpn doesnt claim to be started until its online, and systemd has a start-timeout and will restart it upon timeout
<aszlig>
although patching stuff to use sd_notify is usually not that hard...
<{^_^}>
nixos/nix#2385 (by graham-at-target, 3 minutes ago, open): fetchGit only works on non-annotated tags
<srhb>
gchristensen: I like how you disguised that as a pull request. ++ :-)
<gchristensen>
oh man that was a mistake from me editing a URL I had in my copy buffer :)
orivej has quit [Ping timeout: 252 seconds]
fpletz has joined #nixos-dev
lassulus has joined #nixos-dev
<niksnut>
bleh, maybe we should move to gitter
<niksnut>
too much irc spam
<gchristensen>
:o
<gchristensen>
where are you seeing spam?
<aminechikhaoui>
hurray for gitter :D
* aminechikhaoui
continues his trolling
<samueldr>
niksnut: if it is private message-based spam, setting yourself as +R should fix this; this is the new default, but if your connection is from before the new default, you wouldn't have it
<samueldr>
(you would use /mode niksnut +R )
<samueldr>
august 21st [14:47:48] [freenode-connect] Due to the persistent ongoing spam, all new connections are being set +R (block messages from unidentified users) and will be scanned for vulnerabilities. This will not harm your computer, and vulnerable hosts will be notified.
<niksnut>
samueldr: ah, cool
__Sander__ has quit [Quit: Konversation terminated!]
* gchristensen
is a docs-writing machine
rsa has quit [Ping timeout: 252 seconds]
<niksnut>
gchristensen: +1
<niksnut>
see, we need gitter for proper emotes like :thumbsup:
<gchristensen>
haha
<Dezgeg>
does it support the parrots?
<gchristensen>
niksnut: ok, I think I'm done with doc PRs for the day ;)
<gchristensen>
niksnut: unless you have edits, of course! let me know! I can also send a PR backporting the docs to the 2.1 maintenance branch if you'd ilke.
<niksnut>
no, I can just cherry-pick that
<niksnut>
Dezgeg: I see that Unicode just added a parrot
<samueldr>
I first checked-in with vcunat for the game plan and we agreed it was probably for the best
<samueldr>
gives a little wiggle room for those last PRs, solid date, also gives saturday for those that are burdened by non-nix-related work :)
<domenkozar>
sounds good :)
<domenkozar>
can't wait
<domenkozar>
elm without 3GB closure
<domenkozar>
only took 2 years!
<gchristensen>
:O :D
<domenkozar>
things that excite me, shrimp ready
<domenkozar>
eh, shrink*
<domenkozar>
:D
<gchristensen>
ready shrimp excites me too
<domenkozar>
yeah, just had the best octopus in portugal
<domenkozar>
so it's hard to not think about sea food :D
<domenkozar>
oh jellyfish!
<domenkozar>
damn
<domenkozar>
I lost you
<gchristensen>
hm?
<domenkozar>
nvm :)
timokau[m] has joined #nixos-dev
ckauhaus|away is now known as ckauhaus
drakonis has quit [Remote host closed the connection]
<ckauhaus>
hi, has anyone an idea why Hydra blocks nixpkgs-unstable jobs for more than a week now?
<gchristensen>
looking, ckauhaus
<gchristensen>
we've been stuck at 55-remaining-jobs "forever" now
<LnL>
yeah, there's something with those jobs
<srhb>
Glad someone's looking at them now. :)
<ckauhaus>
gchristensen: yes please :)
orivej has joined #nixos-dev
phreedom has quit [Ping timeout: 250 seconds]
<gchristensen>
love debugging hydra stuff
<domenkozar>
love?
<gchristensen>
"love"
<LnL>
:p
<gchristensen>
hrm. packet-t2-4 keeled over
phreedom has joined #nixos-dev
disasm has joined #nixos-dev
<gchristensen>
ok I think I sorted it
<srhb>
gchristensen: Hooray! Thank you!
<srhb>
gchristensen: And that unstuck latest-finished from the confusing dual nixos-unstable situation as well, looks like. Great!
phreedom has quit [Ping timeout: 250 seconds]
<samueldr>
srhb: AFAIUI the `latest-finished` can still point to an older build, but at least the fix should keep nixpkgs-channels and nixos.org/channels in sync
<srhb>
samueldr: That's what I meant. :) The pointer got bumped, so *presumably* the next update will fix both of the targets :)
ckauhaus is now known as ckauhaus|away
drakonis_ has joined #nixos-dev
<gchristensen>
NixOS's ofborg infrastructure is undergoing a bit of a risky update. Hopefully nothing goes wrong. I'll keep you posted. See #nixos-borg for finer grained updates. Things _will_ be broken for at least a few minutes, but hopefully _only_ a few minutes.
phreedom has joined #nixos-dev
orivej has quit [Ping timeout: 245 seconds]
<gchristensen>
OfBorg is back! Thank you for your patience!
drakonis_ has quit [Remote host closed the connection]