<FRidh>
flokli: many jobs were cancelled because we have several mass rebuilds ongoing and they were not needed anymore
<flokli>
FRidh: yeah, but this build was from a recent evaluation on master
<flokli>
which was the same drv as soemthing previously scheduled, but cancelled on staging-next
<FRidh>
yes that's like you've said, cancellations in one evaluation affect other evaluations
<FRidh>
jobs are cancelled, not jobs in an evaluation
<flokli>
what are jobs in an evaluation?
<flokli>
so, what I assumed here was the following: a) there has been a evaluation for staging-next done, yielding to a build of pipewire to be scheduled b) all builds from that evaluation were cancelled c) another evaluation for nixos:trunk-combined:tested was triggered d) hydra realized it got to the same build of pipewire, but considered it a cached failure
<flokli>
FRidh: does that make sense?
<FRidh>
flokli: yes
<flokli>
well then this is a serious hydra bug
<FRidh>
when you cancel, you cancel a job. However, preferably, you would not cancel a job but only the job in that evaluation. That means if that job is used in another evaluation it would stay alive.
<flokli>
FRidh: wait, I meant c) wasn't yet started when b) happened
<flokli>
running c) will automatically reschedule all builds cancelled by b)?
<FRidh>
yes, I understand you, but it likely is because of the same simplication that was made in the design
<flokli>
I assumed b) "polluted" the nix derivation until manual restart of the build of that individual derivation.
<FRidh>
although actually, I do recall that when you create a new evaluation in a *different* project, it says everything needs to be built even though it was already built in an evaluation in another project
<flokli>
FRidh: but then you still don't know if hydra tries to be smart and "cache the result from a previous build" ("aborted by user") in our case
justanotheruser has quit [Ping timeout: 272 seconds]
<srhb>
domenkozar[m]: In the marketing team paragraph: "If you want to join, you *will* find" -- otherwise lgtm
betawaffle has joined #nixos-dev
<worldofpeace>
domenkozar: "The marketing team was founded and had their first meeting" they've had several, I believe, now
<worldofpeace>
domenkozar: "Erase your darlings" you say "Graham talks about", you should use his full name because not everyone knows him well enough by firstname
<worldofpeace>
everything else seems fine
teto has joined #nixos-dev
<domenkozar[m]>
thanks!
<domenkozar[m]>
I use only first name on purpose :)
justanotheruser has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos-dev
justanotheruser has quit [Ping timeout: 265 seconds]
justanotheruser has joined #nixos-dev
betawaffle is now known as otherwaffle
betawaffle has joined #nixos-dev
otherwaffle has left #nixos-dev [#nixos-dev]
drakonis has joined #nixos-dev
drakonis has quit [Quit: WeeChat 2.8]
cole-h has joined #nixos-dev
abathur has quit [Quit: abathur]
drakonis has joined #nixos-dev
<drakonis>
so, we have a problem with people typosquatting nixos.org
<drakonis>
nxos.org
<gchristensen>
annoying
<drakonis>
and there's some cryptocoin that's similarly named
<drakonis>
a bitcoin fork called nix
<MichaelRaskin>
For the record, .com is _not_ squatting, it was there first
<gchristensen>
probably not a lot we can do about this
<gchristensen>
MichaelRaskin: what, no warning?
<MichaelRaskin>
ouch, right, sorry.
<cole-h>
Warning: .com is probably NSFW
<gchristensen>
not like anyone is home anyway
<gchristensen>
at work*
<drakonis>
very
<MichaelRaskin>
I am too used to the old days where .com being some kind of porn website was well known
<drakonis>
but nxos.org is being typosquatted by a linux distribution
<drakonis>
nitrux lol
<drakonis>
yet another ubuntu clone
<gchristensen>
drakonis: I take it as a sign of success :)
<MichaelRaskin>
Not even sure it counts as squatting
<MichaelRaskin>
I mean, 4-letter .org
<MichaelRaskin>
Cannot pass this up
<MichaelRaskin>
Don't care what else is similarly spelled
<drakonis>
its fairly close anyways
<drakonis>
mildly inconvenient really
<gchristensen>
they have a nice design, let's borrow
<MichaelRaskin>
A person who typos nxos.org and is ready to consider that's where they were heading was not really going to NixOS.org for any good reason
<MichaelRaskin>
(ten more years later) yep, .com was there first, and nxos.org used that colour scheme first, and …
<adisbladis>
drakonis: That could already be achieved by splitting hydra jobsets
<drakonis>
is there any particular reason why it isn't done by default?
<adisbladis>
I'm gonna defer that answer to someone else :)
<samueldr>
drakonis: we do, it's called nixos-unstable :)
<gchristensen>
I think it is a Good Thing they are together, and would rather they be *more* combined
<drakonis>
hmmm, fair enough
<drakonis>
nixos-unstable is just linux builds?
<samueldr>
if you need only linux, it's quite rare that nixos is stuck while nixpkgs continues working
<samueldr>
yes
<drakonis>
hm, neat.
<samueldr>
and since it ends up testing nixos, in addition to building linux things, it may be a bit more solid than nixpkgs-unstable
<drakonis>
very nice.
<cole-h>
Does overriding a derivation and setting `doCheck = false;` force a rebuild?
<samueldr>
cole-h: does changing an input of a derivation force a rebuil? yes, and this. AFAIK will end up changing the environment variables inputs
<cole-h>
Fair enough.
<samueldr>
though it does require the override actually change something
* cole-h
wishes there was a way to disable checkPhase for when things aren't cached, without forcing rebuilds on things that are cached
<cole-h>
I needed libreoffice yesterday. It's not cached due to blocked channel, so it took ~2 hours to build. `doCheck = false;` would have probably sped it up by a significant margin.
* LnL
looks
<jtojnar>
doCheck disables checkInputs so it is potentially breaking
<{^_^}>
nixops#1323 (by adisbladis, 2 hours ago, open): Use rsync instead of scp for file copying
<emily>
+1, scp is a slow mess compared to rsync (though I don't use nixops)
<drakonis>
scp got dunked in favour of basically anything else after ssh became a thing
<gchristensen>
but scp uses ssh?
<emily>
so does rsync
<gchristensen>
that was a reply to drakonis
<emily>
oh
<MichaelRaskin>
cole-h: I would look up the last succesful build on Hydra and just nix-store -r the closure
<gchristensen>
(it literally spawns something to the effect of `ssh remote scp -t` or so -- adisbladis++ for that research)
<{^_^}>
adisbladis's karma got increased to 54
<emily>
scp is basically the most naive possible file transfer algo on top of ssh
<drakonis>
hm, i see
<emily>
and has awful perf/reliability properties
<gchristensen>
anyway, it is less about the "is rsync better?" but more like "is this going to break your expectations of how nixops' upload / download works?"
<drakonis>
ic
<gchristensen>
or "are there traps we're not seeing?"
<FRidh>
are there machines out there, managed with nixops, that don't have rsync installed
<gchristensen>
FRidh: I don't think so -- rsync is in the minimal , mandatory list of packages
<LnL>
adding propagatedUserEnvPkgs = [ rsync ]; should make it install that along with nixops in systemPackages / nix-env
<gchristensen>
nice idea -- could you add a comment to the PR for that?
<LnL>
shlex.join also looked a bit weird at first, but it's for ssh flags
<cole-h>
MichaelRaskin: Oh, I had no idea you could do that. Do I point it to the `Output store paths` or `Derivation store path`?
<MichaelRaskin>
Output store path
<cole-h>
I love you. ✨ MichaelRaskin
<{^_^}>
MichaelRaskin's karma got increased to 32
FRidh has quit [Ping timeout: 264 seconds]
FRidh has joined #nixos-dev
<MichaelRaskin>
More #-dev type question: This trick lookup-latest-on-Hydra is neatly scriptable and easy to use in a way that is not worth it. On the other hand, it is sometimes useful. Do we even have documentation «using Hydra to understand what is in binary cache»?
<MichaelRaskin>
(do we have a natural place for it?)
<gchristensen>
that would be cool and good, MichaelRaskin
<MichaelRaskin>
I have negative amount of clue how it should be targeted and positioned.
<MichaelRaskin>
Negative, because I have no clue re: this about some parts of the main manuals, and here I have even less
<aminechikhaoui>
Hey I have a problem after switching to 20.03, so I have this setting in my nix configuration where I set nix.sandboxPaths = [ "/usr/bin" (toString pkgs.coreutils)]; (for reasons :) )
<aminechikhaoui>
this simple expression http://ix.io/2jo8 would fail to find /usr/bin/env if I run it using nixpkgs from 19.03
<aminechikhaoui>
and it would succeed if I run it with nixpkgs 20.03
<aminechikhaoui>
any idea why, I thought this is strictly build sandbox setup that bind-mount's the paths in the sandbox which seems to be happening
<aminechikhaoui>
oh wait maybe this is a glibc version issue
<aminechikhaoui>
since my /usr/bin/env is now using glibc 2.30 from nixos 20.03
<LnL>
heh, did the impurity just bite you?
* LnL
votes for getting rid of /bin/sh in the sandbox
<aminechikhaoui>
still trying to confirm my assumption though
<aminechikhaoui>
but I can't :D
peelz has quit [Remote host closed the connection]
peelz has joined #nixos-dev
dongcarl has quit [Quit: Ping timeout (120 seconds)]
<aminechikhaoui>
oh doing nix-build test.nix --option extra-sandbox-paths '/usr/bin/env=/nix/store/9v78r3afqy9xn9zwdj9wfys6sk3vc01d-coreutils-8.31/bin/env' seems to fix it
<aminechikhaoui>
so I just need to do "/usr/bin/env=${pkgs.coreutils}/bin/env" in my config
<makefu>
Hey all, is there some documentation to require certain packages to build before the channel advances?I am sure there is some list in nixpkgs. And if so, who decides if the package is worthy?
dongcarl has joined #nixos-dev
<aminechikhaoui>
it's weird though because the config is generated out of pkgs.coreutils and points to extra-sandbox-paths = /nix/store/9v78r3afqy9xn9zwdj9wfys6sk3vc01d-coreutils-8.31 /usr/bin and that hash is glibc 2.27 but my /usr/bin/env is pointing to a different coreutils hash with glibc 2.30
<domenkozar[m]>
srk: spring?
<aminechikhaoui>
oh ~/.config/nix/nix.conf is the one that made me crazy :D
<aminechikhaoui>
I have a hardcoded value of coreutils there for some reason
<aminechikhaoui>
ok mystery solved => PEBCAK
<domenkozar[m]>
hehe
<domenkozar[m]>
impurities considered harmful :D
<aminechikhaoui>
yeah :D
<srk>
domenkozar[m]: okey!
<domenkozar[m]>
yaay
fadenb has joined #nixos-dev
FRidh has quit [Quit: Konversation terminated!]
<hexa->
makefu: iirc you're looked for the tested jobset
<hexa->
see nixos/releae-combined.nix
<makefu>
what are the criteria for adding a package there?
<hexa->
well, I see php in there, so the bar can't be that high :p
<samueldr>
uh?
<samueldr>
php is something that is in use for server type tasks, if it fails it would probably cause a bunch of issues for non desktop users :)
<hexa->
in which case alot more stuff would need to be in there
<hexa->
things like babeld, which when it fails breaks my network :)
<hexa->
makefu: this is about tests completing though, so maybe write a test for an application that depends on scipy and fails rn?
<samueldr>
it's not about everyone's setup, but about covering a good chunk at once AFAIUI
<makefu>
hexa-: for example i want binwalk to be a blessed package. it currently has a checkphase
<hexa->
tough :)
<hexa->
most peoples systems probably don't depend on an interactive application
<hexa->
gnuradio would be another consumer of scipy, but sdr is likely in the same league
fadenb has quit [Quit: killed]
fadenb has joined #nixos-dev
<Ox4A6F>
binwalk is in shell.nix@mobile-nixos. It would be nice to include it somehow in the hydra jobset.
<samueldr>
Ox4A6F: I could, but it wouldn't be part of the tested job
<samueldr>
though that was added for development purposes, assumed to run on a non-aarch64 machine
<worldofpeace>
I think it would be a good thing to backport Jan Tojnar, I guess I could do this.
<worldofpeace>
Another thing I noticed whenever I test gnome in a vm, it appears -vga is forced to "std" and not "virtio". this makes it not a pretty resolution
<jtojnar>
✨️ worlworldofpeace
<jtojnar>
✨️ worldofpeace
<Profpatsch>
-ldofpeace
<Profpatsch>
libdofpeace.so
<worldofpeace>
Profpatsch: the perfect shared object, the more you share the better for the world 😺
<worldofpeace>
Jan Tojnar: can gnome update.sh be made to update to only 3.34, for example
<worldofpeace>
* update.nix
<gchristensen>
reminds me of libiberty
thonkpod has quit [Quit: WeeChat 2.6]
thonkpod has joined #nixos-dev
<jtojnar>
worldofpeace yeah, it is pretty easy – I think we even had it in the past
<jtojnar>
the worst issue is figuring out the API
<domenkozar[m]>
hmm, I managed to get Nix to loop :D
<domenkozar[m]>
while substituting/building
<domenkozar[m]>
on macos
<jtojnar>
worldofpeace I think I came up with something, will test
__monty__ has quit [Quit: leaving]
<worldofpeace>
Jan Tojnar: ooh great. I've currently just thought to backport patches from gnome-3-34 branch to 20.03, gnome-3-28 for pantheon.
<Profpatsch>
> It is the author’s opinion that GitHub actually has no real strategy or plan for interacting with the projects they host, and that they have no plan for continuity when they are as the service provider asked to take actions only a service provider can take. This is incredibly frustrating, and led to a massive loss of trust from Void leadership in the reliability and long-term viability of GitHub as a
<{^_^}>
error: syntax error, unexpected $undefined, expecting ')', at (string):306:17
<cole-h>
emily: ofborg was recently redeployed, so you should be able to build stuff on macOS now :)
<gchristensen>
Profpatsch: I'm not sure if void's source was hosted anywhere else that they'd have been transfered. an IRC channel is pretty different I think. it isn't up to github to provide continuity
<gchristensen>
if you want metadata I can get you metadata
<MichaelRaskin>
Why not actually keep a damned ikiwiki in-tree?
<Profpatsch>
That should be mappable to and from mailinglists and even Github issues/PRs, but somebody has to write it first.
<MichaelRaskin>
Well, we don't have anything that can be honestly called a mailing list anyway
Bunogi1 has joined #nixos-dev
<drakonis>
discourse aint something we could resonably call a mailing list
Bunogi has quit [Ping timeout: 256 seconds]
Bunogi1 is now known as Bunogi
<Profpatsch>
it’s a dag, dawg
<MichaelRaskin>
DAG does not prevent a thing from being faithfully synced with a mailing list
<Profpatsch>
yep, mails can have multiple parents (even though the spec is a mess and every client goes haywire)
<Profpatsch>
I wonder if you can produce cycles by preparing and sending multiple mails at the same time …
<MichaelRaskin>
I think most clients pick a reference that doesn't have descendants in the list of references, and call it a day. They won't even notice you did something fishy. A loop might indeed catch them off-guard
<MichaelRaskin>
I think you can make a reference to a future email with little difficulty, no need to do weird tricks in parallel
Bunogi has quit [Ping timeout: 240 seconds]
justanotheruser has joined #nixos-dev
Bunogi has joined #nixos-dev
<emily>
cole-h: yay!
justanotheruser has quit [Ping timeout: 256 seconds]
<cole-h>
I think we froze on Feb 17th (when the ZHF issue was posted), so it makes sense that the former isn't in 20.03, while the latter is
<cole-h>
Or not froze, but branched?
<worldofpeace>
Branch and feature freeze actually happened before that, hydra was in a serious mood 😁
<cole-h>
Oh, heh
<worldofpeace>
hmm, I think we might need a check that only happens on release branches. somehow checking that packages on the release branch are not newer than on master. almost did that to libsecret