worldofpeace changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | NixOS 20.09 Nightingale ✨ | | | 20.09 RMs: worldofpeace, jonringer |
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-dev
<andi-> We have a ton of 3d old builds on hydra that seem to be "finished" when you click on them (e.g. Could someone restart the queue-runner? It is appears to be clogging up some build slots.
<gchristensen> that is peculiar
rajivr has joined #nixos-dev
tilpner_ has joined #nixos-dev
cole-h has quit [Ping timeout: 240 seconds]
tilpner has quit [Ping timeout: 256 seconds]
tilpner_ is now known as tilpner
orivej has quit [Ping timeout: 265 seconds]
BaughnLogBot has quit [Quit: ZNC 1.6.2+deb1 -]
BaughnLogBot has joined #nixos-dev
BaughnLogBot has quit [Client Quit]
BaughnLogBot has joined #nixos-dev
BaughnLogBot has quit [Client Quit]
pbb has quit [Remote host closed the connection]
pbb has joined #nixos-dev
aminechikhaoui1 has joined #nixos-dev
bgamari_ has joined #nixos-dev
aminechikhaoui has quit [Ping timeout: 256 seconds]
aminechikhaoui1 is now known as aminechikhaoui
bgamari has quit [Ping timeout: 272 seconds]
AlwaysLivid has quit [Remote host closed the connection]
AlwaysLivid has joined #nixos-dev
cole-h has joined #nixos-dev
jonringer has joined #nixos-dev
tilpner_ has joined #nixos-dev
tilpner has quit [Ping timeout: 264 seconds]
tilpner_ is now known as tilpner
pmy has quit [Ping timeout: 256 seconds]
pmy has joined #nixos-dev
jonringer has quit [Ping timeout: 264 seconds]
<siraben> Nice, `nix build .#pkgsCross.sheevaplug.ed` fetches from the cache for me on macOS.
<siraben> So I can properly check cross-compilation and garbage-collect without fear
<siraben> though I can't run the binaries lol
saschagrunert has joined #nixos-dev
<siraben> Ok there's musl64 as well which would work
lukegb has quit [Quit: ~~lukegb out~~]
lukegb has joined #nixos-dev
krkini has joined #nixos-dev
kini has quit [Ping timeout: 260 seconds]
krkini has quit [Quit: bye]
kini has joined #nixos-dev
Gaelan_ is now known as Gaelan
orivej has joined #nixos-dev
rajivr has quit [Quit: Connection closed for inactivity]
__monty__ has joined #nixos-dev
<siraben> example of how moving unzip to nativeBuildInputs fixes cross-compilation:
<symphorien[m]> you can test without cross compiling by setting `strictDeps=true`
<siraben> Why isn't it true by default?
<siraben> Hm, I set it `strictDeps = true;` in the package expression but that didn't seem to do anything
cole-h has quit [Ping timeout: 256 seconds]
<symphorien[m]> <siraben "Why isn't it true by default?"> Someone should take the time to fix the fallout
<symphorien[m]> Notably wrapGAppsHook is incompatible with it
<siraben> How is it supposed to work? I don't see Nix stopping my build if I put cmake in buildInputs
<symphorien[m]> I'm surprised
<siraben> nix (Nix) 2.4pre20210126_f15f0b8 on darwin
<siraben> I'm also considering a `"CC=cc"` → `"CC=${}cc"` treewide to help with cross-compilation
<symphorien[m]> It's not a nix feature but a mkDerivation one
AlwaysLivid has quit [Ping timeout: 268 seconds]
<siraben> nix eval
<siraben> oops wrong window :P
<symphorien[m]> <siraben "How is it supposed to work? I do"> `with import <nixpkgs> {}; stdenv.mkDerivation { name = "foo"; src = sl.src; strictDeps = true; buildInputs = [ cmake ]; buildPhase = "cmake --help"; }` fails with `cmake: command not found` for me
<symphorien[m]> (on linux)
tilpner_ has joined #nixos-dev
tilpner has quit [Ping timeout: 246 seconds]
tilpner_ is now known as tilpner
<sterni> rmcgibbo[m]: does your bot check the targeted branch? it just occurred to me that it should ignore anything targeting haskell-updates since in that case packages that are fixed are still marked as broken until the next set update
<siraben> <siraben "I'm also considering a `"CC=cc"`"> #113295
<{^_^}> (by siraben, 19 minutes ago, open): treewide: add targetPrefix to hardcoded references to CC=cc
<siraben> heh barely any rebuilds
<siraben> symphorien: i'll try that
srk has quit [Remote host closed the connection]
srk has joined #nixos-dev
kini has quit [Remote host closed the connection]
kini has joined #nixos-dev
<siraben> Is there a version of this graph over a longer timespan?
<siraben> symphorien: Ok I can replicate that. `cmake: command not found`
rajivr has joined #nixos-dev
<sterni> siraben: love how you see the times when unstable advances
<clever> symphorien[m]: is there a way to make strictDeps the default? (at the cost of a mass-rebuild)?
<symphorien[m]> probably in the code of stdenv, idk
<clever> siraben: on the subject of changing CC tree-wide, ive found a problem recently (in pico-sdk)
<clever> siraben: the build system assumes $CC is for the host, and will guess target cross-gcc on its own
<clever> so if you use the proper cross stdenv, with CC pointing to cross, the cmake phase fails
<clever> and reading make-derivation.nix, i can see that there is no simple way to turn it on (such as config.nix), so i would have to modify it
<siraben> clever: do you want me to revert the change for pico-sdk?
<clever> no, having CC point to cross is much better
<siraben> oh ok
<clever> it lets me write makefiles that just assume $CC is target, and then i can cross-compile to everything at once
<clever> the cmake needs to be fixed, to not fail like that
<clever> but its an issue you may need to be aware of
<clever> armv6l-linux native is still broken
<clever> and armv6l-linux cross depends on the native gcc
<{^_^}> #107386 (by Gaelan, 8 weeks ago, open): Can't build linux-headers on armv6l
jonringer has joined #nixos-dev
rj has joined #nixos-dev
<rmcgibbo[m]> sterni: Yes, it only builds PRs targeted at master.
<sterni> rmcgibbo[m]: perfect, just saw the haskellPackages PR it built was targeted at master by accident
thibm has quit [Quit: WeeChat 2.9]
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<sterni> rmcgibbo[m]: could you get me a build log for ocamlPackages.uunf on aarch64 perchance?
<sterni> cropping up in a few build failures on aarch64, but the log is truncated too much
<sterni> nvm I can probably use ofborg
<rmcgibbo[m]> sterni: is that in one of the recent PRs and the bot missed the log?
<nbp> niksnut: Did you know Merkle tree before implementing Nix?
<sterni> rmcgibbo[m]: one of the recent reviews, but the issue is the truncated log nixpkgs-review outputs (last 10 lines)
<sterni> rmcgibbo[m]: but don't bother digging it up I think ofborg will work as well :)
<rmcgibbo[m]> oh i see. maybe i can make those print more...
<rmcgibbo[m]> i probably deleted the vm already anyways.
<rmcgibbo[m]> sterni: I've made a note for myself. I'll try to improve that.
<eyJhb> fzakaria: Aha! Ping regarding, can I make it ignore a dependency? Ie. I want it to ignore `org.spigotmc:minecraft-server:jar:1.16.5-SNAPSHOT`, as I `inject` this myself.
<eyJhb> Don't see any way to exclude such a thing. :)
<hexa-> at least 1/3 of the haskellPackages list as "Running builds" on staging-next/x86_64-darwin are already finished :/
<niksnut> nbp: no
<gchristensen> hexa-: I think if you reload the page you'll notice they're not stuck on them, they're just being marked in the db and pass t hrough quickly
<hexa-> hm
<hexa-> might
<hexa-> be some caching, pretty sure I did reload ._.
tilpner has quit [Quit: tilpner]
tilpner has joined #nixos-dev
srk has quit [Remote host closed the connection]
srk has joined #nixos-dev
<siraben> is SDL2 cross-compilation broken on darwin? I have an error about `wayland` not being supported on darwin as the cause of failure
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-dev
saschagrunert has quit [Remote host closed the connection]
rajivr has quit [Quit: Connection closed for inactivity]
<clever> gchristensen: what happened to the old testlog reports in the nixos test framework? they seem to be missing in the python version
<samueldr> clever: the .xml files?
<samueldr> they were removed from the old perl one too
* samueldr digs for discussion
<{^_^}> #87191 (by edolstra, 40 weeks ago, merged): testing{-python}.nix: Remove log pretty-printing cruft
tilpner has quit [Quit: tilpner]
tilpner has joined #nixos-dev
<clever> samueldr: the full html report, with the ability to fold logs, and embed screenshots
<samueldr> yeah
<samueldr> (it was xml)
<clever> > This also makes VM builds more reproducible because the log output contained a lot of timing-sensitive data.
<{^_^}> error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):476:1
<clever> ah, that explains it
<samueldr> not really imo
<samueldr> explains the motivation, but I'm not sure the motivation is valid as it is
<clever> samueldr: i was using the screenshot feature here, because the logs are unreadable without it
<samueldr> haha
<samueldr> the test results log are the real _output_ of the test runs imo
<clever> yeah
<clever> machine # PiXPEX Ei niintiitailailsiisnign gd edveivciecse.s.....oko
<clever> have fun debugging this! :D
<samueldr> two machines running concurrently?
<clever> single machine
<samueldr> how does it even do that?
<clever> not sure
<clever> this is how it boots
<samueldr> hmm, only thing I could see is if it had two serials on stdout?
<samueldr> but doesn't explain why once stage-1 starts it's fine
<clever> i'm thinking its the vga console plus serial console
<clever> and stage1 only prints to one of them
<clever> while ipxe doesnt know where you are, and prints to all
<clever> the perl tester did the same thing
<samueldr> oh, yes, vga console, I actually forgot what was the non-serial one that qemu can do
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-dev
cole-h has joined #nixos-dev
<qyliss> adisbladis: did you investigate why the change in
srk has quit [Remote host closed the connection]
srk has joined #nixos-dev
<ajs124> qyliss: doesn't that happen every time somebody "republishes" a git tag?
<qyliss> if that's so, my question is why it was republished
<qyliss> and what changed
xwvvvvwx has quit [Ping timeout: 240 seconds]
xwvvvvwx has joined #nixos-dev
rj has quit [Ping timeout: 268 seconds]
rj has joined #nixos-dev
<aristid> was there a big purge of nixpkgs committers?
<aristid> i got an email that i lost this privilege :)
<copumpkin> I did too, which is understandable :P
<aristid> if you replied to my second statement, you must’ve typed very quickly
<samueldr> I *assume* it's a run of RFC 55
<copumpkin> aristid: I did :P
<{^_^}> #88867 (by Mic92, 38 weeks ago, open): RFC 55 - Remove inactive Nixpkgs
<samueldr> note that y'all are welcome back
<samueldr> (I assume, still)
<gchristensen> not sure about copumpkin, though ...
* copumpkin looks around shiftily
<copumpkin> >_>
* copumpkin retreats back into the shadows
<tilpner> Yes, of course, this is just to reduce the amount of unused credentials laying around in possibly forgotten places
<aristid> it’s fine, I’m surprised that I didn’t commit last year - does it count PRs? or maybe I really didn’t contribute anything except money last year :p
<tilpner> It should
<samueldr> NixOS is not pay to win!
<samueldr> (but 2019 was just last year!)
<aristid> indeed, it’s march 2020 now right?
<lovesegfault> zimbatm: are you around?
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos-dev
<adisbladis> qyliss: A 1b corruption it seems like
<adisbladis> I did investigate and ran diffoscope on it, I can run it again
<gchristensen> wat
<adisbladis> qyliss:
<qyliss> oh no
<qyliss> it's that thing
<gchristensen> oof
<qyliss> this is the worst git misfeature
<adisbladis> oof indeed
<sterni> rmcgibbo[m]: any way you can detect if a build OOMs for the bot?
<rmcgibbo[m]> Did it happen again?
<rmcgibbo[m]> I've been looking at BPF stuff exactly for that:
<sterni> rmcgibbo[m]: had this failure with your bot and ofborg which looks exactly like another reported failure related to OOM on an Raspberry Pi
<{^_^}> dbuenzli/uunf#15 (by sternenseemann, 6 hours ago, open): Build fails on aarch64 with OCaml 4.08
rj has quit [Quit: rj]
<sterni> (see the linked PRs for the failures occuring in the bot)
LnL- is now known as LnL
<rmcgibbo[m]> Yeah OOM is entirely plausible. I'm logged into a aarch64 c6g instance right now w/o too much memory -- can you give me a cmdline to try? I can run the build and then check dmesg for OOM messages
<supersandro2000> is there a git pre commit hook for nixpkgs?
<supersandro2000> I want to automatically prefix my package changes
<supersandro2000> if not I am writting one
evanjs has quit [Ping timeout: 240 seconds]
evanjs has joined #nixos-dev
<sterni> rmcgibbo[m]: nix-build -A ocamlPackages.uunf '<nixpkgs>'
rj has joined #nixos-dev
rj has quit [Client Quit]
rj has joined #nixos-dev
<rmcgibbo[m]> :( unfortunately it built fine
rj has quit [Remote host closed the connection]
rj has joined #nixos-dev
<rmcgibbo[m]> it's not the OOM killer, unfortunately.
<rmcgibbo[m]> sterni: I don't see memory usage above 10%
<sterni> rmcgibbo[m]: is it inconsistent or does it fail every time?
<sterni> unfortunately I don't really know what exited with 2 may mean as it is not documented for ocamlopt
<rmcgibbo[m]> sterni: same spot, ~4 times in a row
<sterni> alright weird
<sterni> wait did you use ocamlPackages? I thought it was an issue on 4.08 exclusively so far :|
<sterni> seems like it affects 4.10 as well then
<rmcgibbo[m]> I used `nix build -f . ocamlPackages.uunf jackline ocamlPackages.carton` from current nixpkgs HEAD as my cmdline
<sterni> ohh but I think the 4.10 one succeeds
<sterni> jackline is 4.08
<sterni> maybe it's just a weird compiler bug
mkaito has joined #nixos-dev
mkaito has joined #nixos-dev
mkaito has quit [Changing host]
<hexa-> can someone restart
<hexa-> it has one failing test which I can't repro locally
rj has quit [Quit: rj]
rj has joined #nixos-dev
<andi-> restarted
<hexa-> thx
<eyJhb> supersandro2000: did you ever find the place/something which stated the `doCheck = false;`?
<eyJhb> Anyways this is ready for merge -
<{^_^}> #112237 (by eyJhb, 1 week ago, open): pythonPackages.transmission-rpc: 0.11 -> 3.2.2
__monty__ has quit [Quit: leaving]
<supersandro2000> eyJhb: wdym?
<supersandro2000> also left you two comments
<supersandro2000> mmmmhh
<supersandro2000> it should work ™️
<eyJhb> supersandro2000: updated to reflect comments
<rmcgibbo[m]> Aha! I've figured out how to use BPF to get notified whenever any process gets an ENOSPC or the kernel OOM killer gets called.
<rmcgibbo[m]> Linux is wild.
<V> woah
<V> that's cool
<rmcgibbo[m]> r-rmcgibbo will hopefully be getting a bit smarter soon (not posting build failures under these conditions)
tilpner_ has joined #nixos-dev
tilpner has quit [Ping timeout: 240 seconds]
tilpner_ is now known as tilpner