worldofpeace changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | NixOS 20.09 Nightingale ✨ https://discourse.nixos.org/t/nixos-20-09-release/9668 | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html | https://r13y.com | 20.09 RMs: worldofpeace, jonringer | https://logs.nix.samueldr.com/nixos-dev
<infinisil> Would be nice if Nix printed updates while deleting garbage
<infinisil> Currently it just says "deleting '/nix/store/trash'", which can take hours without any extra output
<gchristensen> you mean the last step?
<infinisil> Yea
<gchristensen> yea
<gchristensen> sudo strace -e unlink -fp ...
<gchristensen> but only run it for a second and ctrl-c it
<infinisil> Neat, why ctrl-c?
<samueldr> nix has to delete *all* files, so it's quite slow; though I wonder if it could print once per top-level thing it's deleting, or some progress status thing
<gchristensen> strace slows it down significantly
<infinisil> I'd love some "X out of Y paths deleted"
<samueldr> that'd be nice
<infinisil> This seems to be the function doing the deletion: https://github.com/NixOS/nix/blob/master/src/libutil/util.cc#L386
<infinisil> Eh, don't feel like digging this late
<infinisil> Well, more than I already did anyways
<cole-h> :P
rajivr has joined #nixos-dev
<infinisil> I think the best way to monitor a GC is to look at the filesystem usage
<infinisil> zfs list, shows that my store only shrank from 121GB to 114GB, so there's still a bunch to go..
<infinisil> Though this also makes me feel like I'm missing some obvious optimization which would delete files that much quicker
<cole-h> dd if=/dev/zero of=/dev/sd*
<infinisil> Hmm, what if I create a new /nix dataset, copy over *only* the things that would not be GCd, then destroy the original /nix
<infinisil> I expect this GC to delete like 80% of space. So just copying the 20% and doing a dataset destroy might be faster
<gchristensen> sounds slower than a million unlinks
<infinisil> From https://serverfault.com/questions/801074/delete-10m-files-from-zfs-effectively I get the notion that ZFS is particularly slow at deleting things
<infinisil> On the other hand, maybe I should just GC more often than once a year :P
<gchristensen> lol
<gchristensen> you could learn about filesystems, kernels, zfs, performance counters, etc. and try to learn about tuning ZFS to delete faster
<ajs124> hm. something's weird. when I locally build "nix-build nixos/release.nix -A netboot.x86_64-linux" on release-20.09, $out/nix-support/hydra-build-products exists and contains the right data, but https://hydra.nixos.org/build/129492856 has an empty nix-support folder
<gchristensen> lassulus found that earlier I think
<ajs124> how does something like that happen? is hydra up to something strange? it seems to keep the build-products for the iso jobs.
<gchristensen> maybe something w.r.t. symlinks?
ris has quit [Ping timeout: 240 seconds]
<ajs124> hm, I don't see any symlinks anywhere
<ajs124> but https://github.com/NixOS/hydra/pull/818/commits/9cc76f6d69aa69ff47697f3b7698f8d95f7dbbac touched build-result.cc, so maybe something in there?
<gchristensen> all the artifacts in the netboot build are symlinks, is what I'm thinking
<ajs124> hm, yeah. nix-support/hydra-build-products references the actualy store paths, not the symlinks, though.
<ajs124> that might be a relevant difference, actually. for the iso, it's all contained in the same store path, for netboot it's not.
<gchristensen> yeah, that is what I meant -- but didn't explain very well :') I'm glad you see that as interesting too
<ajs124> also, neither of the two tests in hydra cover this. I'll implement one and see if it fails.
<andi-> ajs124++
<{^_^}> ajs124's karma got increased to 0o7
<dstzd> sdasdasdas
<dstzd> sorry. wrong window
<cole-h> Which window is the right window to type that? 🤔 😂
<gchristensen> ajs124++
<{^_^}> ajs124's karma got increased to 8
<dstzd> pm. helping someone with their irc notification script. lol
<cole-h> Heh
<infinisil> 423439 store paths deleted, 168638.54 MiB freed
<infinisil> Compression is awesome. these paths only used ~100GB
<gchristensen> nice
puck has quit [Read error: Connection reset by peer]
puckipedia has joined #nixos-dev
puckipedia is now known as puck
alp has joined #nixos-dev
justan0theruser has joined #nixos-dev
justanotheruser has quit [Ping timeout: 240 seconds]
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nixos-dev
<siraben> 5403 store paths deleted, 10549.09 MiB freed
<siraben> I GC pretty often, especially since working on a cross-compilation toolchain takes up a lot of space, heh.
alp has quit [Ping timeout: 268 seconds]
justan0theruser has quit [Ping timeout: 268 seconds]
justanotheruser has joined #nixos-dev
alp has joined #nixos-dev
cole-h has quit [Ping timeout: 256 seconds]
jonringer has quit [Ping timeout: 260 seconds]
orivej has joined #nixos-dev
alp has quit [Ping timeout: 264 seconds]
orivej has quit [Ping timeout: 260 seconds]
kalbasit has quit [Ping timeout: 240 seconds]
saschagrunert has joined #nixos-dev
alp has joined #nixos-dev
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nixos-dev
codyopel has quit [Quit: Idle for 30+ days]
luc65r has joined #nixos-dev
__monty__ has joined #nixos-dev
Baughn has quit [Ping timeout: 260 seconds]
Baughn has joined #nixos-dev
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nixos-dev
orivej has joined #nixos-dev
FRidh has joined #nixos-dev
alp has quit [Remote host closed the connection]
alp has joined #nixos-dev
<andi-> What is the current status around nix on the new Apple Arm platform? Does that work yet?
<LnL> thefloweringash has some WIP stuff
<LnL> and the 10.11 changes are being tested on hydra, but hydra is going through the queue a bit slowly
<thefloweringash> I have a cross compiled set of bootstrap tools, an implementation of ad-hoc code signing, and a natively built hello using those bootstrap tools.
<thefloweringash> the catch is that it relies a lot on apple's sdk for header files and library definitions
alp has quit [Ping timeout: 240 seconds]
alp has joined #nixos-dev
<{^_^}> firing: RootPartitionLowInodes: https://status.nixos.org/prometheus/alerts
<siraben> TIL about git worktree which is amazing
<{^_^}> resolved: RootPartitionLowInodes: https://status.nixos.org/prometheus/alerts
<siraben> clever: is there an update/alternative to https://github.com/cleverca22/not-os ?
<clever> siraben: its mostly as an example of where to start, when building your own thing
<siraben> clever: not sure if you saw my ping yesterday about https://github.com/NixOS/nixpkgs/pull/102503
<{^_^}> #102503 (by siraben, 18 hours ago, open): Initial implementation of remarkable1 cross-compile
<siraben> clever: I see. I haven't been able to get not-os building on darwin, even with the Linux builder
<clever> what error did it give?
<siraben> Ideally I would want to generate .iso images like Tiny Core Linux-type distributinos
<siraben> clever: this error http://ix.io/2CS8
<clever> thats due to using too new of a nixpkgs version
<clever> try an older nixpkgs
<siraben> How old?
<clever> plan a: check the git history to see when systemd.user was added to that file
<siraben> I'll try mid 2019
<clever> you can see it was working on 2020-08-14, but then it broke on 2020-08-24
<clever> and a git rev is right there on each one
<siraben> I see.
<siraben> So the remarkable toolchain works however the maintainer would like me build GCC from source instead of using the company provided toolchain
<siraben> There's a community toolchain that achieves the same goal; https://github.com/toltec-dev/toolchain
alp has quit [Ping timeout: 272 seconds]
<siraben> clever: once question i have is how do you deal with the long iteration times when working on adding a cross toolchain to nixpkgs?
<siraben> Should I set up a cachix for that?
<clever> cachix just stores the builds, but you can just store them locally
<clever> nix-store -r /nix/store/foo --add-root bar --indirect
<clever> that will create an indirect root (just like the result nix-build makes)
<clever> and then foo cant be deleted by GC'd
<siraben> I see.
<siraben> Also, how do I override lib.systems.examples to add a new system in an overlay?
<clever> also, i fixed the eval on master
<clever> you need to use -I nixpkgs=/path/to/nixpkgs, to use the fork where you added the new system
alp has joined #nixos-dev
<siraben> Thanks, will try it out.
FRidh has quit [Quit: Konversation terminated!]
<siraben> clever: something like this; https://github.com/siraben/nix-remarkable/
<clever> and yeah, "use an older nixpkgs" kinda conflicts with "use a recent fork with your cross"
<clever> the eval now passed on both master and unstable
<gchristensen> manveru: where did nix-inclusive go?
<siraben> clever: it built, how do I run it in qemu?
<clever> siraben: there should be a qemu script already in not-os
<clever> base.nix: exec ${pkgs.qemu_kvm}/bin/qemu-kvm -name not-os -m 512 \
<clever> yeah, you want to build `-A system.build.runvm`
<clever> and it will generate a script that runs qemu with the right args for you
<siraben> nix build --argstr system 'x86_64-linux' -f release.nix system.build.runvm
<siraben> error: attribute 'system' in selection path 'system.build.runvm' not found
alp has quit [Ping timeout: 272 seconds]
<siraben> clever: i tried with other nix files, doesn't seem to work
<clever> its not part of release.nix though
<clever> default.nix should have it
<clever> [clever@amd-nixos:~/not-os]$ nix-build -A config.system.build.runvm
<manveru> gchristensen: we moved it to iohk
<gchristensen> oh
<gchristensen> the move didn't create a redirect, so I started referencing my fork instead
<gchristensen> I'll move back, thanks
<manveru> sorry about that... i basically just tried to inverse the fork relationship :)
<manveru> but github doesn't make that easy
<gchristensen> no worries
<gchristensen> has anyone noticed a severe slowdown in nix-copy-closure or other copying? I'm seeing 2-3s per `.drv` copied
<clever> ive looked into it before, and there is no way to undo a fork and make it a self-contained repo
<clever> the official docs recommend you just delete the whole thing, then re-make it
<siraben> clever: qemu-system-x86_64: Invalid parameter 'vlan'
<clever> siraben: you may need to edit some of those -net options, or just delete them, in base.nix
<clever> its not been tested in a while
<siraben> Ooh edited, looks like it's booting!
<siraben> clever: are there similar projects like not-os that produce isos?
<clever> siraben: does it need to boot from a cd, or usb?
<siraben> USB
<siraben> clever: seems to get stuck, http://ix.io/2CSw
<clever> siraben: you would need a modified form of this, that copies the files to the /boot and runs grub-install
<clever> siraben: i dont think not-os presents a shell by default, it just runs the services that are configured
<supersandro2000> the darwin machine we where talking yesterday
<clever> siraben: but ssh is one of the services it runs
<supersandro2000> I got the machine and I am currently installing
<siraben> clever: I see.
<siraben> What about the message `3 Nov 13:29:43 ntpdate[114]: no servers can be used, exiting`?
<clever> siraben: ntpdate failed to set the clock
<clever> runit.nix: ${pkgs.ntp}/bin/ntpdate 192.168.2.1
<supersandro2000> I need to follow this page https://github.com/NixOS/ofborg/wiki/Operating-a-Builder to get the builder ready?
<clever> siraben: i hard-coded it to the ip of my router, oops!
<siraben> Haha
<siraben> And which IP do I SSH into?
<clever> siraben: that depends on the -net flags you used when qemu ran it
<clever> the qemu docs can help with that
<siraben> I see. And why does not-os only compile on linux?
<clever> darwin->linux cross-compiling isnt tested as much
<clever> and not-os defaults to trying to build for the current host
<clever> so it would try and make a darwin binary that runs under the linux kernel, and then fail to boot
<clever> this is another project i did recently, even more stripped down then not-os
<clever> the initrd contains busybox, a few select kernel modules, and a tiny .c binary to measure boottime
<clever> then a shell script (line 42) configures things
<siraben> I see. Could this compile in darwin as well?
<clever> not sure
<clever> its setup to cross-compile to arm
<matthewkenigsber> Is there a good reason python packages should depend on bash? Like the python wrapper uses bash to set a few env variables, but that's all I've seen. That could be done in Python without the extra dependency
costrouc has joined #nixos-dev
costrouc has quit [Quit: costrouc]
costrouc has joined #nixos-dev
<niksnut> matthewkenigsber: yes, ideally we would have some notion of an "intrusive" makeWrapper that sets env vars *inside* the program it's wrapping (for supported scripting languages)
alp has joined #nixos-dev
<matthewkenigsber> niksnut: that would get rid of an unnecessary fork as well, right?
<niksnut> an unnecessary exec
<matthewkenigsber> Thanks
<aminechikhaoui> Hey niksnut, any chance this tiny change can be merged https://github.com/NixOS/nix/pull/4116 ? I've been using it locally for few weeks and didn't notice any issues
<{^_^}> nix#4116 (by AmineChikhaoui, 3 weeks ago, open): support name attribute in fetchGit
costrouc has quit [Quit: costrouc]
costrouc has joined #nixos-dev
<siraben> clever: how can I take the output of not-os and turn it into a disk image to flash?
<siraben> It's already a .img but I'm not sure what to do next.
misuzu_ has joined #nixos-dev
colemickens has quit [*.net *.split]
domenkozar[m] has quit [*.net *.split]
ryantm has quit [*.net *.split]
cstrahan has quit [*.net *.split]
flokli has quit [*.net *.split]
worldofpeace has quit [*.net *.split]
nh2[m] has quit [*.net *.split]
JJJollyjim has quit [*.net *.split]
rnhmjoj has quit [*.net *.split]
Ericson2314 has quit [*.net *.split]
thefloweringash has quit [*.net *.split]
Dandellion has quit [*.net *.split]
ajs124 has quit [*.net *.split]
regnat has quit [*.net *.split]
misuzu has quit [*.net *.split]
pikajude has quit [*.net *.split]
worldofpeace has joined #nixos-dev
Ericson2314 has joined #nixos-dev
Dandellion has joined #nixos-dev
thefloweringash has joined #nixos-dev
JJJollyjim has joined #nixos-dev
regnat has joined #nixos-dev
rnhmjoj has joined #nixos-dev
cstrahan has joined #nixos-dev
ryantm has joined #nixos-dev
domenkozar[m] has joined #nixos-dev
nh2[m] has joined #nixos-dev
pikajude has joined #nixos-dev
flokli has joined #nixos-dev
ajs124 has joined #nixos-dev
flokli has quit [Max SendQ exceeded]
Ericson2314 has quit [Max SendQ exceeded]
alp has quit [Ping timeout: 272 seconds]
maralorn has quit [Ping timeout: 244 seconds]
bennofs[m] has quit [Ping timeout: 244 seconds]
michaelpj has quit [Ping timeout: 246 seconds]
ryantm has quit [Ping timeout: 246 seconds]
worldofpeace has quit [Ping timeout: 246 seconds]
JJJollyjim has quit [Ping timeout: 246 seconds]
jonge[m] has quit [Ping timeout: 240 seconds]
jtojnar has quit [Ping timeout: 270 seconds]
kraem has quit [Ping timeout: 244 seconds]
siraben has quit [Ping timeout: 270 seconds]
domenkozar[m] has quit [Ping timeout: 246 seconds]
Dandellion has quit [Ping timeout: 246 seconds]
thefloweringash has quit [Ping timeout: 246 seconds]
aanderse has quit [Ping timeout: 244 seconds]
Valodim[m] has quit [Ping timeout: 244 seconds]
bbigras has quit [Ping timeout: 272 seconds]
aterius has quit [Ping timeout: 246 seconds]
Ox4A6F has quit [Ping timeout: 246 seconds]
regnat has quit [Ping timeout: 246 seconds]
nh2[m] has quit [Ping timeout: 246 seconds]
rnhmjoj has quit [Ping timeout: 246 seconds]
Irenes[m] has quit [Ping timeout: 272 seconds]
philipp[m] has quit [Ping timeout: 240 seconds]
DamienCassou has quit [Ping timeout: 244 seconds]
treed[m] has quit [Ping timeout: 244 seconds]
timokau[m] has quit [Ping timeout: 244 seconds]
danielrf[m] has quit [Ping timeout: 244 seconds]
matthewkenigsber has quit [Ping timeout: 246 seconds]
bk1603[m]1 has quit [Ping timeout: 244 seconds]
puzzlewolf has quit [Ping timeout: 244 seconds]
emily has quit [Ping timeout: 244 seconds]
roberth has quit [Ping timeout: 260 seconds]
rycee has quit [Ping timeout: 260 seconds]
ma27[m] has quit [Ping timeout: 260 seconds]
alexarice[m] has quit [Ping timeout: 272 seconds]
flokli has joined #nixos-dev
<Taneb> I've hit an error running hydra-backfill-ids for my work's Hydra server
<Taneb> I get this error: main::backfillJobsJobsetId(): DBI Exception: DBD::Pg::st execute failed: ERROR: duplicate key value violates unique constraint "jobs_pkey"
kraem has joined #nixos-dev
Valodim[m] has joined #nixos-dev
bennofs[m] has joined #nixos-dev
siraben has joined #nixos-dev
michaelpj has joined #nixos-dev
aanderse has joined #nixos-dev
maralorn has joined #nixos-dev
jtojnar has joined #nixos-dev
ryantm has joined #nixos-dev
worldofpeace has joined #nixos-dev
jonringer has joined #nixos-dev
aterius has joined #nixos-dev
bbigras has joined #nixos-dev
jonge[m] has joined #nixos-dev
domenkozar[m] has joined #nixos-dev
thefloweringash has joined #nixos-dev
Dandellion has joined #nixos-dev
nh2[m] has joined #nixos-dev
rnhmjoj has joined #nixos-dev
Ox4A6F has joined #nixos-dev
regnat has joined #nixos-dev
Irenes[m] has joined #nixos-dev
colemickens has joined #nixos-dev
kraem has quit [Quit: Bridge terminating on SIGTERM]
aanderse has quit [Quit: Bridge terminating on SIGTERM]
michaelpj has quit [Quit: Bridge terminating on SIGTERM]
Valodim[m] has quit [Quit: Bridge terminating on SIGTERM]
siraben has quit [Quit: Bridge terminating on SIGTERM]
maralorn has quit [Quit: Bridge terminating on SIGTERM]
bennofs[m] has quit [Quit: Bridge terminating on SIGTERM]
jtojnar has quit [Quit: Bridge terminating on SIGTERM]
ryantm has quit [Quit: Bridge terminating on SIGTERM]
jonge[m] has quit [Quit: Bridge terminating on SIGTERM]
aterius has quit [Quit: Bridge terminating on SIGTERM]
bbigras has quit [Quit: Bridge terminating on SIGTERM]
Dandellion has quit [Quit: Bridge terminating on SIGTERM]
domenkozar[m] has quit [Quit: Bridge terminating on SIGTERM]
regnat has quit [Client Quit]
nh2[m] has quit [Quit: Bridge terminating on SIGTERM]
rnhmjoj has quit [Quit: Bridge terminating on SIGTERM]
Ox4A6F has quit [Quit: Bridge terminating on SIGTERM]
colemickens has quit [Client Quit]
worldofpeace has quit [Quit: Bridge terminating on SIGTERM]
thefloweringash has quit [Quit: Bridge terminating on SIGTERM]
Irenes[m] has quit [Client Quit]
jtojnar has joined #nixos-dev
alp has joined #nixos-dev
aterius has joined #nixos-dev
Irenes[m] has joined #nixos-dev
bennofs[m] has joined #nixos-dev
Ox4A6F has joined #nixos-dev
kraem has joined #nixos-dev
JJJollyjim has joined #nixos-dev
timokau[m] has joined #nixos-dev
thefloweringash has joined #nixos-dev
colemickens has joined #nixos-dev
DamienCassou has joined #nixos-dev
Valodim[m] has joined #nixos-dev
danielrf[m] has joined #nixos-dev
aanderse has joined #nixos-dev
philipp[m] has joined #nixos-dev
Ericson2314 has joined #nixos-dev
Dandellion has joined #nixos-dev
alexarice[m] has joined #nixos-dev
jonge[m] has joined #nixos-dev
ryantm has joined #nixos-dev
emily has joined #nixos-dev
maralorn has joined #nixos-dev
roberth has joined #nixos-dev
regnat has joined #nixos-dev
michaelpj has joined #nixos-dev
nh2[m] has joined #nixos-dev
rycee has joined #nixos-dev
domenkozar[m] has joined #nixos-dev
bbigras has joined #nixos-dev
mkg20001 has joined #nixos-dev
siraben has joined #nixos-dev
matthewkenigsber has joined #nixos-dev
worldofpeace has joined #nixos-dev
treed[m] has joined #nixos-dev
ma27[m] has joined #nixos-dev
bk1603[m] has joined #nixos-dev
puzzlewolf has joined #nixos-dev
rnhmjoj has joined #nixos-dev
zowoq[m] has joined #nixos-dev
FRidh has joined #nixos-dev
saschagrunert has quit [Remote host closed the connection]
saschagrunert has joined #nixos-dev
cole-h has joined #nixos-dev
<__red__> Potentially silly question. When we compile for say arm, in our arm build environment - that's going to create a different result than say cross-compiling to arm
<__red__> Can nix cross-compile packages to other architectures?
<__red__> ie: Can I get an amd64 installation of nix to build an arm boot image
<__red__> or arm nixos installation cd?
<sorear> you can cross-compile packages, see https://nixos.org/manual/nixpkgs/stable/#chap-cross . I haven't personally cross-compiled an entire boot image
<srhb> Is bastion having memory issues?
<srhb> I guess it's got a lot of swap and is just being slow...
<__red__> thanks
<__red__> I'm guessing the cross-compilation builds aren't going to be binary identical to the natively compiled - even with the samer versions
<__red__> Hmm - thanks for the reference - I'll take a peek!
<sorear> I think "binary identical" is a good goal but there's no way to enforce that in a single build
<cole-h> It's been almost 8 hours since "update-all-channels.timer" last fired -- is there anything to be done about that?
<cole-h> (I assume that's the reason nixos-unstable hasn't updated yet, despite a successful eval+build completed ~8 hours ago
<cole-h> )
<sorear> you could set up qemu (boot.binfmt.emulatedSystems, configuration.nix) and do a "native" build, which would be a bit more likely to be the same (expect a 5-10x slowdown from emulation)
<srhb> cole-h: That was the reason I asked about memory, before I realized about swap.
<srhb> cole-h: The service seems to be "activating" for a long while, and then it isn't.
<cole-h> right
* srhb shrugs
<__red__> Wow
<__red__> "It Just Works"
alp has quit [Ping timeout: 246 seconds]
<supersandro2000> gchristensen LnL Who do I need to ask to get the darwin machine attached to ofborg? Should I create an issue?
misuzu_ has quit [Quit: leaving]
misuzu has joined #nixos-dev
ris has joined #nixos-dev
<FRidh> is there a way to have substituteAllInplace fail when a replacement cannot be made?
alp has joined #nixos-dev
<supersandro2000> wrap the command in if command; then true; else false; fi ?
alp has quit [Remote host closed the connection]
alp has joined #nixos-dev
<samueldr> supersandro2000: that won't work substituteAllInPlace returns a success when it doesn't substitute anything
<samueldr> I would suggest not proposing solutions when it's only a supposition
<supersandro2000> most linux tools return something not 0 if they fail
<FRidh> where to draw the line regarding backwards-incompatible changes in Nixpkgs
<FRidh> I am changing the wrapping of Python packages, so the wrapping is separated from propagating builds. That is, you don't get all these other non-Python packages in your wrappers that happened to be propagated
<FRidh> fixing this throughout is some work, but doable
<samueldr> maybe describe *what* the incompatibility means?
<FRidh> next step, is to get rid of the propagation alltogether, or at least, do it differently
<gchristensen> I'm seeing a lot of problems with nixUnstable, even the version just backported. a lot of random mysterious failures with remote builds, deadlocks in the client, unreleased store path locks, deadlocks in the daemon which ultimately get killed after the system reaches shutdown.target, etc.
<gchristensen> anyone seeing these?
<supersandro2000> Do python packages rely on binaries in dependencies normally?
<FRidh> in that first step, it means writing `requiredPythonModules` instead of `propagatedBuildInputs`
<supersandro2000> I am having a lot of troubles with remote builds getting killed but I am not sure if it is cause if nix
<FRidh> gchristensen: not using remote builds, but i keep having issues with running out of file descriptors
<samueldr> FRidh: if you use propagatedBuildInputs it would fail silently, right? is there a way to make it fail loudly?
<samueldr> or would there still be reasons to propagate a python package?
<FRidh> samueldr: I cannot think of a way, you will really only notice during runtime that your wrappers aren't complete anymore
<samueldr> right, the worst kind of backward incompatibilities, those that break things silently :/
<FRidh> indeed
<supersandro2000> only had file descriptors running out with nix when I used watchman
<samueldr> I was thinking that python "modules" (do we have a terminology for that?) could somehow export something through passthru or $out/nix-support that tells the python builders they're python modules so that the builder could break things, FRidh
<FRidh> anyway, now step 2) is to stop using `propagatedBuildInputs` altogether for Python packages.
<FRidh> samueldr: exactly! That's what I am implementing now. a) using requiredPythonModules in nix, as much as possible. And b) using $out/nix-support/required-python-modules to build up $PYTHONPATH during a build
<samueldr> could the python builders bail when a propagatedBuildInput includes that?
<samueldr> if so, you have an annoying thing to deal with to fix builds, but magnitudes better than silently failing
<samueldr> silently not-failing-but-not-working!
<FRidh> yes, but I think it should still be possiblre to propagate python applications
<FRidh> right
<samueldr> yeah, that was what I was asking first
<FRidh> b), creating PYTHONPATH, is done by Python's setup hook. However, I am of the opinion that packages should not have hooks. Instead, hooks should refer to the package. In that case, Python packages would only propagate the hook, but not the interpreter. That will solve certain issues when you have multiple interpreters
<gchristensen> sigh. recursive nix is deadlocking now too. fool's errand.
<FRidh> but for b), it would mean replacing `python` with `pythonHook` everywhere...
<FRidh> or at least, many places
zarco has joined #nixos-dev
<FRidh> maybe actually not that many, given python packages will propagate the hook
<FRidh> hmm
<samueldr> how common is it to really want to propagate a python package?
<samueldr> aren't propagated dependencies already quite frowned upon?
<FRidh> true
<FRidh> in that sense, I don't think python should be any different here
<samueldr> maybe you could have the builder opt-in in that case
<samueldr> yeah, I would agree if it wasn't for the problem on hand :)
<samueldr> doPropagatePythonDependencies = true; # defaults to false
<FRidh> if the builder can opt-in, it means making a change already
<FRidh> at which point you may as well make the actual fix
<samueldr> maybe it's hard to do the fix, or maybe they only want to get the damn thing going
<samueldr> and later fix it
<FRidh> yes its not much of a problem to keep support for that for a release or so
<FRidh> that would reduce the impact of step 1) at least
<FRidh> for reference, here is what I am working on https://github.com/NixOS/nixpkgs/pull/102613
<{^_^}> #102613 (by FRidh, 8 hours ago, open): WIP: Python: wrap and patch using `requiredPythonModules` instead of `propagatedBuildInputs`
<cole-h> gchristensen: Maybe you should use the <details> tags on https://github.com/NixOS/nix/issues/4216... :P
<{^_^}> nix#4216 (by grahamc, 13 minutes ago, open): Recursive Nix deadlocks often
<cole-h> Only slighly long without them ;)
<gchristensen> I forgot how to make them :)
saschagrunert has quit [Remote host closed the connection]
kalbasit has joined #nixos-dev
FRidh has quit [Quit: Konversation terminated!]
alp has quit [Ping timeout: 272 seconds]
alp has joined #nixos-dev
<gchristensen> sigh, the strict validation of hostName has broken the standard behavior for OSes on Packet and that means it has to be "weird"
<gchristensen> and actually there is no correct way for me to handle it
<cole-h> (View raw)
<cole-h> That's how you hide it :P
<ehmry> this might be offtopic, but did nixos ever have much of a self-hosted phase before github? asking for a former gentoo user
<gchristensen> yes
<ajs124> the commit history tells me it used to be on svn
<gchristensen> and it used to be on this strange bug tracker
<gchristensen> like lemongrass or something
<ehmry> I should read up on the history of distros, I should know more about this
<samueldr> according to archive.org it was a JIRA instance https://web.archive.org/web/20080514141938/http://bugs.strategoxt.org/browse/NIX
<samueldr> (followed the bug tracker link https://web.archive.org/web/20080425213923/http://nixos.org/ )
<samueldr> ah!
<samueldr> with something else in-between
<samueldr> ehmry: ^ no real details about the svn installation though AFAICT
<samueldr> the URL from 2008 seems to indicate some "repoman" thing
<samueldr> and hopefully the history of NixOS (org and project) is never off-topic here :)
* ehmry remembers why we all move to git+github
<ehmry> I'm getting flashbacks of using older bugtrackers
<gchristensen> yup
<gchristensen> ah yellowgrass
<supersandro2000> How do you do those star things?
<domenkozar[m]> evening
<ehmry> hello
<ehmry> I don't like stars, but I'm realizing that my dislike of stars has made me forget how much of an improvement github was
<gchristensen> yup
__monty__ has quit [Quit: leaving]
ChanServ has quit [shutting down]
ChanServ has joined #nixos-dev
luc65r has quit [Quit: WeeChat 2.9]
<edef> really wish good issue trackers were readily available
<edef> i don't think github issues are very good either tbh
justanotheruser is now known as mrdulus
<supersandro2000> > * ehmry remembers why we all move to git+github
<supersandro2000> I mean this
<{^_^}> error: syntax error, unexpected '*', at (string):345:1
<supersandro2000> * mmm
mrdulus is now known as justan0theruser
* supersandro2000 waiting for someone to have time for the darwin ofborg machine
rajivr has quit [Quit: Connection closed for inactivity]
supersandro2000 has quit [Read error: Connection reset by peer]
supersandro2000 has joined #nixos-dev
alp has quit [Read error: Connection reset by peer]
alp has joined #nixos-dev
jonringer has quit [Remote host closed the connection]
Baughn has quit [Ping timeout: 256 seconds]
Baughn has joined #nixos-dev
<roberth> could someone cancel https://hydra.nixos.org/eval/1624198? It's a big one but has a broken golang
<roberth> gchristensen: ^ you helped me out last time <3
<worldofpeace> roberth: I cancelled all scheduled builds
<worldofpeace> it takes a bit though and I only just clicked it :D
<roberth> thanks, worldofpeace++
<{^_^}> worldofpeace's karma got increased to 257
<roberth> oh sorry I ruined your 2^8
<worldofpeace> oh right 2^8 was cool. it was gonna get bigger anyways 🤣
<worldofpeace> roberth: okay 16322 builds have been cancelled.
<roberth> new eval will have to run some of those, but still 5326 saved, thanks!