taktoa has quit [Remote host closed the connection]
contrapumpkin is now known as BTC_Boar
BTC_Boar is now known as contrapumpkin
orivej has joined #nixos-dev
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm has joined #nixos-dev
<disasm> gchristensen: if terraform only tracks hardware-configuration, how would you go about with patching? For example, glibc update where hardware-configuration.nix stays the same. Terraform would think the state is fine and wouldn't run the nixops parts.
<grahamc> disasm (IRC): in this model terraform would not ever call nixops and once the Infra was up nixops would takeover
<disasm> oh, I misread it the opposite where you ran terraform to execute nixops
<disasm> I thought provider "nixops" actually ran nixops
<grahamc> Oh I see. No. It is more like just registering the node in the nixops “database”
<disasm> cool :)
<grahamc> There is a chef_node resource which is similar. It sets the node up in chefs database but doesn’t run chef.
<disasm> I've also been poking around with this: https://github.com/awakesecurity/nix-deploy Although nixops feels more powerful.
<grahamc> Bed time :) I’ll be thinking more on this Monday
<disasm> yeah, I'm very excited :)
<disasm> Happy to help write some of the code as well or review :)
<disasm> ooh... this fixes the abstraction of hardware as well!
<disasm> so you could define different hardware on an instance in terraform, hardware-configuration is auto-generated, then nixops will just use that hardware defined in terraform!
<grahamc> Yes!! Yes! Exactly!
lopsided98 has quit [Remote host closed the connection]
lopsided98 has joined #nixos-dev
lopsided98 has quit [Remote host closed the connection]
<clever> the way i configure zfs, is to auto-generate a unique hostid via /dev/urandom
<clever> so every install has a unique id, that nixops must be aware of when building the images
<clever> zfs also adds many options for what datasets to mount where, which nixops has to know of, and it currently lacks the ability to create them on the fly
lopsided98 has joined #nixos-dev
<clever> and if your in a non-zfs mood, that matters more, because the partitions are even more locked up
stqism has quit [Quit: Like 3 fire emojis lit rn 🔥🔥🔥]
stqism has joined #nixos-dev
zarel has joined #nixos-dev
FRidh2 has joined #nixos-dev
orivej has quit [Ping timeout: 264 seconds]
JosW has joined #nixos-dev
zarel has quit [Ping timeout: 248 seconds]
ma27 has joined #nixos-dev
ma27 has quit [Ping timeout: 276 seconds]
ma27 has joined #nixos-dev
ma27 has quit [Ping timeout: 248 seconds]
ma27 has joined #nixos-dev
vcunat has joined #nixos-dev
FRidh2 has quit [Quit: Konversation terminated!]
FRidh2 has joined #nixos-dev
orivej has joined #nixos-dev
zarel has joined #nixos-dev
ma27 has quit [Ping timeout: 256 seconds]
zarel has quit [Read error: Connection reset by peer]
zarel_ has joined #nixos-dev
zarel_ has quit [Ping timeout: 276 seconds]
orivej has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos-dev
zarel has joined #nixos-dev
zarel has quit [Quit: Leaving]
vcunat has quit [Ping timeout: 256 seconds]
vcunat has joined #nixos-dev
orivej has joined #nixos-dev
ma27 has quit [Ping timeout: 276 seconds]
jtojnar has quit [Remote host closed the connection]
ma27 has joined #nixos-dev
ma27 has quit [Ping timeout: 276 seconds]
<gchristensen> ofborg now schedules builds of the package identified as changed in the commit messages: https://github.com/NixOS/ofborg/tree/next#automatic-building, example: https://github.com/NixOS/nixpkgs/pull/34541#issuecomment-362816933
FRidh2 is now known as FRidh
ma27 has joined #nixos-dev
orivej has quit [Ping timeout: 240 seconds]
<pie_> i hope one day nixos will make this a nonissue https://twitter.com/koszuldude/status/958888114633695232
<pie_> or something.
<MichaelRaskin> Won't change the command-line part, but maybe can make understandable what is safe to do…
alunduil has quit [Ping timeout: 256 seconds]
<pie_> well yeah i dont mean cli is inherently bad or anytihng
<MichaelRaskin> Somehow the person in the thread thinks that GUI installation has any less chance of ending up with multiple python versions installed…
<FRidh> ah with jupyter/ipython it gets more tricky because of multiple entry points to the same thing. I encountered this myself last week, where I had a system-wide installation of a Python env with I think it was our `notebook` package, and then used `nix run` with `python3.pkgs.nbconvert`. `nbconvert can be executed as nbconvert but also as jupyter nbconvert. They don't give the same result.
jtojnar has joined #nixos-dev
<gchristensen> well this is confusing and wrong: https://github.com/NixOS/nixpkgs/pull/34558#event-1456485026
<FRidh> gchristensen: what do you mean?
<gchristensen> well it shouldn't have labeled that PR with python
<gchristensen> and I don't know why it did :P
__Sander__ has joined #nixos-dev
ma27 has quit [Ping timeout: 256 seconds]
<FRidh> gchristensen: heh I know that feeling. Somehow I seem to be the only one that can actually build the following without error nix-build https://github.com/NixOS/nixpkgs/archive/0186286433cccbf6faf288c22df03e63e46dd94d.tar.gz -A python2.pkgs.odfpy
<gchristensen> O.o
ma27 has joined #nixos-dev
<FRidh> gchristensen: think it does the opposite?
<gchristensen> it seems to be surprising behavior in how git handles `commit..commit` in `git diff --name-only`
<FRidh> oh you are not using the API for this?
<gchristensen> no, because I already have it checked out
<gchristensen> I could, I suppose, but I'm not :)
<gchristensen> hrm. the auto-builds could get _real_ annoying...
michaelpj_ has joined #nixos-dev
<MichaelRaskin> Is it news to you?
<MichaelRaskin> I assume that lack of [WIP]-filtering means also lack of Chromium-filtering
<gchristensen> that part doesn't seem so annoying
<MichaelRaskin> Does it build _all_ revdeps?
<gchristensen> I'm not sure what that means
<MichaelRaskin> What exactly it builds for trusted PRs? Everything that a nox-review pr would (i.e. a lot)?
<gchristensen> oh, no, only valid attributes identified in the top-line of the commit message: https://github.com/NixOS/ofborg/tree/next#automatic-building
__Sander__ has quit [Ping timeout: 240 seconds]
<MichaelRaskin> Oh wait, not everyone is subscribed to all notifications and has a separate mailbox for NixpkgsÖ
<MichaelRaskin> Yes, for some people it could increase the amount of notifications quite a bit
<FRidh> could indeed generate a lot of messages on https://github.com/NixOS/nixpkgs/pull/34553
<gchristensen> when it triggers the build, it treats it as a single build instruction, ie: `build awscli spyder` which will cause a single message per architecture
<gchristensen> for example, expand the Partial Log on https://github.com/NixOS/nixpkgs/pull/34541#issuecomment-362818726 -- all three of the identified packages are built together
<FRidh> having 75 commits that could each have a valid attribute in their commit name could then generate up to 75*3 messages, right?
<MichaelRaskin> You mean 75 commits in a single PR?
<FRidh> yes
<gchristensen> if they are pushed individually to the PR, then it would, but if the PR is opened and already has 75 commits it would only trigger 1*3
<pie_> ok i dont know how i did this but a nix script build was failing, then i did a collect garbage, and then it wasnt failing anymore
<pie_> and it wasnt a disk space issue
<gchristensen> FRidh: does that make sense?
<FRidh> gchristensen: yes, it does
ma27 has quit [Ping timeout: 276 seconds]
<gchristensen> if it turns out to be noisy, I don't mind turning it off until we find a better solution or comment coalescing is implemented (https://github.com/NixOS/ofborg/issues/47)
<vcunat> Hmm, is it possible to just edit the comment that called borg and put the result in there?
<vcunat> (there's a permission issue, I guess)
<FRidh> but having a way to disable it (e.g. marked as work in progress) would be nice. E.g., with larger PR's where I update the python package set I expect failures and do not intend to fix everything either.
<MichaelRaskin> Not sure it is generally a good idea to automatically edit human comments…
<gchristensen> there actually isn't, github's permissions are not good. because ofborg has permission to label issues, it can edit anybody's comments too
<gchristensen> FRidh: that does seem good
<vcunat> It would be only comments that specifically asked for it.
<vcunat> (Assuming Borg won't get mad.)
<MichaelRaskin> (looking at Python label for Doxygen PR) that's a bold assumption
<vcunat> :-)
<MichaelRaskin> Also, right now it is allowed to add a borg mention in the end of a substantial comment…
<gchristensen> or in the middle (or top) of a substantial comment, I think editing a person's comment is not good
<MichaelRaskin> Given what I know, I would say mentioning borg at the top is unwise.
<MichaelRaskin> Imagine mentioning borg at the top above a multiparagraph text where one of the paragraphs starts with «Build this in the following order: …» and then goes on to mention Python or LibreOffice
<MichaelRaskin> It stops at newlines?
<gchristensen> yep
<gchristensen> I'd like to talk more about excluding WIP PRs. could we talk more in #nixos-borg?
<MichaelRaskin> As for me, when I am in #nixos-dev I am almost always also in #nixos-borg
<MichaelRaskin> (Now is not exceptional)
<vcunat> There are now some bad errors on Hydra that I can't reproduce locally: https://hydra.nixos.org/build/68633155
<vcunat> (same output hash) Fortunately I've only seen these on i686-linux so far.
<vcunat> only
<Dezgeg> something related to #!/bin/sh in new nix?
<vcunat> it is some shell thing
<Dezgeg> I noticed something similar-sounding on the aarch64 box
<Dezgeg> it might reproduce there
<vcunat> I'd expect ./configure to be run with stdenv's shell, not /bin/sh.
<Dezgeg> not unless someone calls patchShebangs
<vcunat> Ah, I see, we just call it.
<vcunat> That might be a good stdenv change - to auto-patch the script before running, by default.
ckauhaus has joined #nixos-dev
Dezgeg has quit [Quit: WeeChat 1.0.1]
jtojnar_ has joined #nixos-dev
michaelpj_ has quit [Ping timeout: 256 seconds]
jtojnar has quit [Ping timeout: 256 seconds]
JosW has quit [Quit: Konversation terminated!]
JosW has joined #nixos-dev
Dezgeg has joined #nixos-dev
orivej has joined #nixos-dev
jtojnar_ has quit [Quit: jtojnar_]
jtojnar has joined #nixos-dev
jtojnar_ has joined #nixos-dev
jtojnar has quit [Ping timeout: 264 seconds]
jtojnar_ is now known as jtojnar
FRidh has quit [Quit: Konversation terminated!]
<contrapumpkin> peti: is there an endorsed place to put haskell packages in nixpkgs that aren't on hackage?
<contrapumpkin> I see hie packages in a separate file, but this doesn't fit into there either
vcunat has quit [Quit: Leaving.]
JosW has quit [Quit: Konversation terminated!]
<yorick> domenkozar: we're looking into using stack2nix for hydra builds. is something like that feasible?
ckauhaus has quit [Quit: Leaving.]
<LnL> contrapumpkin: vaultenv isn’t on hackage, you can add it like that
<contrapumpkin> oh interesting, thanks
<dtz> new nix uses busybox basic shell, it' doesn't provide all of the features many derivations expect from /bin/sh
<dtz> see https://github.com/NixOS/nix/issues/1832, I've patched it on my machine and builders to avoid mass failures
<dtz> (I explicitly mention the busybox failure in that issue)
<dtz> *limited busybox shell, works better if we enable required features
yegortimoshenko has quit [Ping timeout: 255 seconds]
yegortimoshenko has joined #nixos-dev
LnL has quit [Quit: exit 1]