<copumpkin>
nothing that would name them according to their position that I know of
takle has quit [(Ping timeout: 255 seconds)]
sigmundv has quit [(Ping timeout: 240 seconds)]
mizu_no_oto has joined #nixos
dustmote has joined #nixos
tnias has quit [(Ping timeout: 260 seconds)]
markus1 has quit [(Ping timeout: 258 seconds)]
dustmote is now known as mooet
mooet has quit [(Client Quit)]
tnias has joined #nixos
dustmote has joined #nixos
markus1 has joined #nixos
mudri has quit [(Quit: WeeChat 1.7)]
dustmote has quit [(Read error: Connection reset by peer)]
dustmote has joined #nixos
dustmote has quit [(Client Quit)]
derjohn_mob has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
filterfish has quit [(Ping timeout: 260 seconds)]
derjohn_mob has quit [(Ping timeout: 240 seconds)]
roconnor has quit [(Ping timeout: 240 seconds)]
dustmote has joined #nixos
dustmote has quit [(Client Quit)]
ndowens08 has joined #nixos
nixy has quit [(Quit: WeeChat 1.5)]
<nh2>
can somebody explain me via what ways nix figures out that it has to use the generic haskell builder when I use mkDerivation on a Haskell package?
ttuegel has quit [(Remote host closed the connection)]
nixy has joined #nixos
dustmote has joined #nixos
takle has joined #nixos
filterfish has joined #nixos
<copumpkin>
shlevy: you around?
takle has quit [(Ping timeout: 260 seconds)]
dustmote has quit [(Quit: bye)]
hamishmack has quit [(Ping timeout: 240 seconds)]
takle has joined #nixos
takle has quit [(Ping timeout: 260 seconds)]
hamishmack has joined #nixos
systemfault has quit [(Quit: Bye!)]
slack1256 has joined #nixos
takle has joined #nixos
filterfish_ has joined #nixos
takle has quit [(Ping timeout: 260 seconds)]
filterfish has quit [(Ping timeout: 268 seconds)]
s33se has joined #nixos
s33se_ has quit [(Ping timeout: 260 seconds)]
mbrgm has quit [(Ping timeout: 255 seconds)]
nh2 has quit [(Ping timeout: 260 seconds)]
mbrgm has joined #nixos
RchrdB has joined #nixos
RchrdB has quit [(Ping timeout: 260 seconds)]
takle has joined #nixos
takle has quit [(Ping timeout: 255 seconds)]
hexagoxel_ has joined #nixos
RchrdB has joined #nixos
Fare has quit [(Ping timeout: 255 seconds)]
systemfault has joined #nixos
hexagoxel has quit [(Ping timeout: 260 seconds)]
hexagoxel_ is now known as hexagoxel
ryanartecona has joined #nixos
ryanartecona has quit [(Read error: Connection reset by peer)]
mizu_no_oto has joined #nixos
RchrdB has quit [(Ping timeout: 255 seconds)]
roconnor has joined #nixos
RchrdB has joined #nixos
slack1256 has quit [(Remote host closed the connection)]
ryanartecona has joined #nixos
ndowens08 has quit [(Ping timeout: 260 seconds)]
ryanartecona has quit [(Read error: Connection reset by peer)]
ryanartecona has joined #nixos
RchrdB has quit [(Ping timeout: 255 seconds)]
systemfault has quit [(Quit: Bye!)]
<copumpkin>
hmm
<copumpkin>
"error: current Nix store schema is version 10, but I only support 7"
RchrdB has joined #nixos
<copumpkin>
I don't recall doing anything that would upgrade the nix store schema
<copumpkin>
oh dammit
<copumpkin>
I was developing on Nix master on this machine
<copumpkin>
and it must've silently upgraded the store
<tnks>
a skip from 7 to 10 sounds pretty severe.
<Sonarpulse>
tnks: 7-10 is indeed the case
<tnks>
how many months between those commits?
<Sonarpulse>
copumpkin: I'm sorry
<Sonarpulse>
I guess I did a forced downgrade
<Sonarpulse>
on nixos with reinstall
<tnks>
also, how do I check what version my /nix/store is at?
<Sonarpulse>
I cannot recall whether I lost items in store
dustmote has joined #nixos
<Sonarpulse>
but I suppose it wasn't too bad
<copumpkin>
I'm just going to run --dump-db on the new one and --load-db on old one
<copumpkin>
I think that should work
<Sonarpulse>
oh didn't know about those
<Sonarpulse>
when I say reinstall
<Sonarpulse>
I mean nixos flash drive and nixos-install over original
<Sonarpulse>
which has never felt too distruptive
<Sonarpulse>
what you proposal is probably better
<copumpkin>
this is on a remote machine
dustmote has quit [(Client Quit)]
<copumpkin>
it's a PITA :)
Fare has joined #nixos
mizu_no_oto has quit [(Ping timeout: 260 seconds)]
mizu_no_oto has joined #nixos
<copumpkin>
yeah, --load-db and --dump-db saved the day
<clever>
Havvy: i think the bigger issue is going to be the git nixpkgs is stored on
<Havvy>
As long as nobody checks in two PDFs with the same SHA-1, shouldn't be an issue.
<Havvy>
But what checksum is used for /nix again?
<tilpner>
Havvy - Well, it does use sha1 hashes everywhere. If someone links colliding file A into nixpkgs, he could later exchange it for colliding file B and nobody would notice
<tilpner>
More collisions would have to be found, but that's apparently not so hard anymore :/
digitalmentat has quit [(Ping timeout: 240 seconds)]
<Havvy>
So, should Nix try moving away from using SHA1?
takle has joined #nixos
<clever>
Havvy: nix will package the entire directory tree up as a NAR, then hash that with sha256 i think
<clever>
maybe 512
<johnw>
Havvy: it depends on whether the ability to create colliding files introduces any attack vectors for a Nix system
dustmote has joined #nixos
<clever>
tilpner, Havvy: but of note with git, all files have a header of "blob %d\0" prepended to them, including the size of the data
<johnw>
it's not like files are going to start colliding all of a sudden, when they didn't before
<Havvy>
clever: That didn't save webkit's git repo.
<clever>
Havvy: i believe they used svn
<Havvy>
Ah, true.
proteusguy has joined #nixos
<simpson>
And SHA-1-based deduplication.
<clever>
git does its object storage and de-dup after prefixing the files with the above mentioned header
<clever>
which destroys any pre-existing collisions
dustmote has quit [(Read error: Connection reset by peer)]
<tilpner>
It sort of does violate determinism, e.g. when re-downloading files and suddenly being served a colliding file with different content. nix won't complain, yet I get a different system. Won't happen by accident, so you'd change from sha1 for security, not determinism
dustmote has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
dustmote has quit [(Client Quit)]
<Heffalump>
clever: but presumably an attacker would construct a collision that included that header (if possible)
<clever>
Heffalump: ive heard something about the attack working by appending large amounts of specialy crafted junk, which changes the size in that header, causing the state of the sha to become scrambled
systemfault has joined #nixos
<Heffalump>
clever: yeah, maybe the current attack won't work, but finding one that deals with that header too (e.g. work out how much junk there'll be in advance) might happen in future
<Heffalump>
where does nix use sha1? I thought sha256 was the standard hash?
<srhb>
I'm not sure how to debug this, but chromium really doesn't consider itself the default browser. If I clear my mimeapps.list, and redo xdg-settings set default-web-browser chromium.desktop, adds text/html=userapp-Nightly-H9BP6X.desktop under Default Applications, but what on earth is this?
<clever>
Heffalump: fixed-output derivations can use any hash, md5, sha1, sha256, sha512, md5 is already being phased out, and sha1 can easily be removed as well
<srhb>
That file doesn't exist in the store or in my profile, and if I grep through my profile, it isn't there either.
<tilpner>
Heffalump - It's not "everywhere", but $ rg 'sha1 = ' ~/dev/nixpkgs | wc -l outputs 8684, so it finds some usage
MercurialAlchemi has joined #nixos
mw has quit [(Quit: kumquat)]
mw has joined #nixos
<tilpner>
(8705 with 'sha1\s+=')
eacameron has quit [(Remote host closed the connection)]
Sonarpulse has quit [(Ping timeout: 260 seconds)]
<srhb>
Hmm, if I remove that file, xdg-settings get default-web-browser still returns userapp-Nightly-...
<srhb>
Where would that information live?
<srhb>
grepping through the entire store might be a bit.. Expensive. :-P
eacameron has joined #nixos
<tilpner>
srhb - Are you looking for the userapp*desktop? Check ~/.local/share/applications
<srhb>
tilpner: I just deleted everything there. xdg-settings get default-web-browser still returns userapp-Nightly-H9BP6X.desktop
<srhb>
I just did grep -R "userapp-Nightly" 2>/dev/null through /nix/store with no results. This is very weird.
<srhb>
find / -iname "userapp-Nightly-H9BP6X.desktop" 2>/dev/null -- no results either.
<srhb>
I'm pretty sure this is some weird Firefox remnant, but I simply don't understand how this information lingers. I don't even have Firefox installed, and I've collected garbage just now
<clever>
srhb: this will show you every single syscall its doing, so you can "grep --color Nightly logfile*"
<srhb>
I don't know why I never reach for this
<srhb>
It's in my "oooh, spooky" arsenal, but I do know how to use it. :-P
<clever>
srhb: as long as its not using mmap to read the offending string, you will see exactly where the data came from
reinzelmann has joined #nixos
<srhb>
And problem solved! Apparently, the values in ~/.config/mimeapps.list silently override the ones in ~/.local/applications, even though xdg-settings set default-web-browser was writing to ~/.local/share/applications/mimeapps.list (albeit the WRONG value.)
<srhb>
xdg-utils is one impressive clusterfuck.
xadi has quit [(Ping timeout: 255 seconds)]
<srhb>
Thanks for the help though!
<clever>
:)
<srhb>
Eck, chromium still doesn't think it's the default browser though...
<clever>
some browsers will want the un-wrapped version of their binary in the config
<srhb>
Darn.
<clever>
but you need to run the nix wrapper to make it start right
<clever>
i just set it right and turn the warning off
<srhb>
Oh right.
<srhb>
So chromium-browser
<srhb>
No, that doesn't work...
<clever>
/nix/store/vn267yxglzz5wfil9ry32mspbs9f3y9x-chromium-55.0.2883.87/bin/chromium is a bash script
<clever>
that runs /nix/store/mprk6qiibik8aixslv4jwgb1a7dnfl62-chromium-55.0.2883.87/libexec/chromium/chromium
<clever>
the bash script adds the configured plugins to env vars, before running the 2nd chromium
<clever>
since they are in different derivations, you can recompile the bash script, and reuse the pre-compiled chromium
<clever>
but now there are 2 chromium binaries, not even counting chromium-browser yet
<srhb>
Sorry, I don't follow. I need to override that shell script to make it believe it's actually the default browser?
qualia has quit [(Remote host closed the connection)]
<clever>
i think the problem is that the actual browser isnt aware of the nixpkgs wrappers
<clever>
and nobody has patched that
<srhb>
I swear this used to work, I just have no idea why
<srhb>
It's pretty obvious, because without that knowledge, chromium will do a first run every time, never restoring old tabs.
<clever>
mine still restores tabs correctly, even if its not default
<srhb>
(And the annoying notification about settings default browser as well...)
<clever>
did you run it with sudo at any point?
<srhb>
I sure hope not.
<srhb>
But I guess I can try nuking its settings from orbit
xadi has joined #nixos
<clever>
or chown them
eacameron has joined #nixos
takle has joined #nixos
<NixOS_GitHub>
[nixpkgs] cstrahan pushed 1 new commit to master: https://git.io/vyInf
<NixOS_GitHub>
nixpkgs/master 1ba97d6 Charles Strahan: ghcjsHEAD: unbreak
<NixOS_GitHub>
[nixpkgs] bjornfor opened pull request #23232: Align attrname with pkgname of some packages (master...align-attrname) https://git.io/vyIcT
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vyIcO
Havvy has quit [(Quit: Computer Restarted or Restarting IRC)]
Havvy has joined #nixos
thc202 has joined #nixos
hotfuzz has joined #nixos
<makefu>
tempeh: if it does not require external dependencies (which is almost never the case) you could use buildPythonPackage only populated with name,version and src. for everything else there is http://nixos.org/nixpkgs/manual/#sec-python
<tempeh>
makefu: right - basically i need exactly buildPythonPackage, but without wheel
<makefu>
is the project not capable to produce a wheel?
justbeingglad has left #nixos []
<tempeh>
makefu: it has only two python dependencies, but it does a lot more setting up in 'setup.py install'. using buildPythonPackage as it is seems to just run 'setup.py bdist_wheel', which doesn't do the other stuff
<tempeh>
it doesn't seem like it
eacameron has joined #nixos
<tempeh>
i know everything builds correctly if i just run 'setup.py install' with a couple of dependencies in the path, id like to just do that
FRidh has joined #nixos
<FRidh>
tempeh: you can use buildPythonPackage but you should disable the buildPhase and override the installPhase with `python setup.py install_data --install-dir=$out --root=$out`
<FRidh>
at least, something like that.
<tempeh>
i see what you're getting at. ill try that thanks
edvorg- has joined #nixos
<FRidh>
tempeh: have a lookt at `$ git grep "setup.py install"` Hopefully some of those 'examples' still function.
eacameron has quit [(Ping timeout: 264 seconds)]
edvorg- is now known as edvorg
edvorg has quit [(Remote host closed the connection)]
edvorg has joined #nixos
jensens has joined #nixos
<tempeh>
FRidh: yep i found a couple examples doing that in. very helpful ty
eacameron has joined #nixos
ilyaigpetrov has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
abcrawf has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vyIWN
<domenkozar>
seems like certificate is not served?
<ikwildrpepper>
getting a 4045
<ikwildrpepper>
404
<domenkozar>
strange, chrome complains it's not secure
<domenkozar>
maybe they added some protection against cloudflare due to recent security problems?
<goibhniu>
Hi, I configured my shellInit like this: http://sprunge.us/APTH to enable the gpg-agent, but I get "gpg-connect-agent: can't connect to the agent: IPC connect call failed" ... is shellInit the wrong place to do that?
<NixOS_GitHub>
[nixpkgs] rickynils opened pull request #23233: nixos/virtualisation/gcp: Use lib/make-disk-image for building Google… (master...gcp-make-image) https://git.io/vyIub
rmeis has quit [(Ping timeout: 260 seconds)]
<NixOS_GitHub>
[nixpkgs] ts468 pushed 1 new commit to master: https://git.io/vyIzP
<NixOS_GitHub>
nixpkgs/master eb073e9 Thomas Strobel: moneyplex: fix: add download link
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 2 new commits to master: https://git.io/vyISn
<NixOS_GitHub>
nixpkgs/master fdd9d7f Joe Hermaszewski: haskell-packages: Add minimal benchDepends support...
<NixOS_GitHub>
nixpkgs/master 14b6f2a Domen Kožar: Merge pull request #23235 from expipiplus1/haskell-bench...
<qknight>
kde5 lacks icons here anyone knows how to install kde5 properly? (i'm using i3 with kde5 apps as konsole and okular)
eacameron has joined #nixos
filterfish_ has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar closed pull request #22797: haskell: support Haskell Program Coverage (HPC) (master...hpc) https://git.io/vDiPW
magnetophon has joined #nixos
d762b440__ has quit [(Read error: Connection reset by peer)]
mizu_no_oto has joined #nixos
derjohn_mob has quit [(Ping timeout: 264 seconds)]
<NixOS_GitHub>
[nixpkgs] domenkozar closed pull request #22378: haskell-modules: split off nix-specific overrides and common ones (master...haskell-split-overrides) https://git.io/vD36w
eacameron has quit [(Ping timeout: 240 seconds)]
<domenkozar>
haskell is ready for 17.03 except one issue :)
rsa_ is now known as rsa
<iMatejC>
qknight: missing package kde5.breeze-icons in systemPackages ? also... output of kde5 app.. lets say: okular, would help to diagnose the missing icons problem
<nh2>
peti: in http://lists.science.uu.nl/pipermail/nix-dev/2016-June/020585.html you say "we'll honor the mandated version bounds for libraries" but further below you say "it's a good idea to use Stackage as a *recommendation* for our package set, but we cannot expect to be 100% compliant to Stackage" which seems to contradict it. What is the current guarantee nixpkgs gives in terms of library versions, are they promised to be the LTS versions 1:1 for l
mkoenig has joined #nixos
joelmo has joined #nixos
<joelmo>
Do someone have NixOS installed alongside with another OS? I have NixOS and want to be able to boot into kubuntu
pi3r has quit [(Ping timeout: 252 seconds)]
<nh2>
joelmo: what's stopping you from doing so? Can't you install the two alongside each other in the same way you'd install any two Linux distributions alongisde each other?
<nh2>
or is your question about how to install two Linux distributions alongside each other in general?
<NixOS_GitHub>
[nixpkgs] globin pushed 3 new commits to master: https://git.io/vyI7F
<NixOS_GitHub>
nixpkgs/master 8d95889 Robin Gloster: xf86_input_multitouch: fix build
<NixOS_GitHub>
nixpkgs/master e001b72 Robin Gloster: wv2: fix build
<NixOS_GitHub>
nixpkgs/master bbefb5d Robin Gloster: warmux: fix build
<joelmo>
Yes how do I do it. Starting a normal installer would maybe overwrite /boot
pi3r has joined #nixos
<sziszi>
qknight: you need to run some kde services to get icons afaik
<sziszi>
this is not nixos specific
<nh2>
joelmo: you'd simply have to tell your installer which partitions to use; the two OSs can use different partitions for their /boot (if they use different partitions for boot at all). For example, in the (K)Ubuntu installer, you can pick that "I want to pick the partitions myself" options instead of "override whole disk" and "install alongside other OS" option, then you can set yourself to which partition what should go
loupgaroublond has quit [(Ping timeout: 256 seconds)]
<nh2>
I have not used nixos's installer yet, so I don't know what that looks like, but since the Ubuntu installer makes this quite easy, I guess an approach I'd recommend would be to first install nixos, an then use the Ubuntu installer, making sure to create separate partitions for your Ubuntu
<sziszi>
one of my friend using this without problems on another system
<sziszi>
nh2: nixos doesn't have graphical installation afaik
<nh2>
sziszi: yes,that's why I suggest to install nixos first, then see what the partitions look like in the Ubuntu installer and make extra partitions appropriately -- that would work even if we don't know what the nixos installer does
loupgaroublond has joined #nixos
<nh2>
joelmo: you can also use `gparted` from a live USB stick to organise your partitions before using any installer, I find that one easier to use than the partitioning tool inside the Ubuntu installer
<nh2>
(gparted is already installed on the live USB mode that contains the Ubuntu installer)
DutchWolfie has joined #nixos
DutchWolfie has quit [(Changing host)]
DutchWolfie has joined #nixos
nschoe has joined #nixos
bachp has quit [(Ping timeout: 240 seconds)]
bachp has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 1 new commit to master: https://git.io/vyI5x
<NixOS_GitHub>
nixpkgs/master 7ed28a0 Domen Kožar: R: fix bio3d build
d762b440__ has joined #nixos
filterfish_ has quit [(Read error: Connection reset by peer)]
griff_ has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<gchristensen>
sometimes when reviewing commits I'd like to be able to skip to the next commit that doesn't mention a thing ex: "Skip until we're out of the ocaml patchset"
d762b440__ has quit [(Read error: Connection reset by peer)]
d762b440__ has joined #nixos
<pierron>
peti: that's a summary of what we discussed at the FOSDEM.
pi3r has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] domenkozar closed pull request #23104: R/generate-shell: Use R from current tree, fix missing wget dep. (master...fix/generate-shell) https://git.io/vyeVv
<NixOS_GitHub>
[nixpkgs] domenkozar closed pull request #19531: r-modules: fix and use nix-shell --pure to generate package lists [wip] (master...rmodules/generate-shell) https://git.io/vPwjn
cpennington has joined #nixos
sheenobu has joined #nixos
Wizek has joined #nixos
Wizek_ has joined #nixos
johann__ has quit [(Quit: Leaving.)]
copumpkin has joined #nixos
takle_ has quit [(Remote host closed the connection)]
<copumpkin>
niksnut: the basic issue is that EC2 images are b0rked right now (at least if you want to configure from userdata)
<copumpkin>
and for now my choices are the old race condition we had before or not working at all :P
<copumpkin>
that's why my systemd service issue came up
<pierron>
domenkozar: you are referring to the SOS rfc?
rmeis_ has joined #nixos
<domenkozar>
pierron: nope
<SuprDewd>
hmm, I have a Nix derivation (mathematica9 to be specific) that defines its own unpackPhase. however, when I go into nix-shell '<nixpkgs>' -A mathematica9, and then run unpackPhase, it doesn't do what the derivation defined. am I misunderstanding something?
<copumpkin>
niksnut: anyway, just trying to avoid getting broken EC2 images into 17.03 :) I guess we can figure it out and fix after branch-off
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vyIxB
<bennofs>
copumpkin: just reading your Nix build-time dependency issue, would you consider pkg in `import ${pkg}` a build-time dependency?
<copumpkin>
yeah
d762b440__ has quit [(Read error: Connection reset by peer)]
<bennofs>
because the path for that will not even appear in the .drv since it is a "eval time dependency"
<copumpkin>
oh, you meant in raw nix
<copumpkin>
sorry, I thought you meant in a string
<bennofs>
yeah in raw nix
<copumpkin>
then nope, we basically don't do IFD in nixpkgs
<copumpkin>
I'd love to be able to account for all of that stuff but not going to set the standards too high :)
<peti>
pierron: Yes, I saw it! Great job, I'm glad you took the time to write it down. I'll offer feedback ASAP.
pi3r has joined #nixos
freusque has joined #nixos
filterfish_ has quit [(Read error: Connection reset by peer)]
d762b440__ has joined #nixos
cfricke has joined #nixos
<NickHu>
What's the difference between build-max-jobs and build-cores? Which should I be seting, and do they affect the hashes of derivations I build?
<copumpkin>
doesn't affect hashes
cfricke has quit [(Client Quit)]
<copumpkin>
former affects number of concurrent nix jobs, latter affects concurrency within each of those jobs
cfricke has joined #nixos
<copumpkin>
so latter translates to make -jN for example, if you have a Makefile-based builder
johann__ has quit [(Quit: Leaving.)]
<copumpkin>
the former is safer, since some builds get weird and don't work well concurrently
<gchristensen>
it is possible that build-cores produces results which are not bit-for-bit identical, or even may have bugs (though this is less likely)
<NickHu>
So I should only set build-max-jobs then?
<gchristensen>
that is what I do, but it probably isn't strictly bad to set build-cores to more than 1
<copumpkin>
yeah, because of the potentially weird behavior you get from more cores, packages have to opt into supporting that
<copumpkin>
which is frustrating sometimes :)
<copumpkin>
but big ones like the linux kernel have it set
<copumpkin>
otherwise you can set --cores all you want and the build will still not pass -j to make (or whichever other build tool is being used)
filterfish_ has joined #nixos
<copumpkin>
globin: so I have a few changes I won't get into 17.03 by the branch-off. How long do I actually have? :D
d762b440__ has quit [(Read error: Connection reset by peer)]
<gchristensen>
12-1pm EST, copumpkin
<copumpkin>
lol I mean I won't get them in by then :)
<gchristensen>
:)
<globin>
copumpkin: what are those changes?
<copumpkin>
I kind of injected a foot-in-the-door I guess by pushing something that I thought worked, but is actually broken
<bennofs>
pierron: I'm reading that RFC. Doesn't the RFC result in many derivations needing to created multiple times? Since each pkg needs to create the derivation for its deps
<copumpkin>
globin: getting the EC2 image to actually work again :)
<gchristensen>
oh you're fine then, it is a bug-fix
<copumpkin>
globin: right now it doesn't spin up new services that you configure at startup. It applies the system configuration properly, but doesn't enable services
<copumpkin>
err, start
<NickHu>
So if my ~/.nixpkgs/config.nix is empty, then build-max-jobs is inherited from /etc/nix/nix.conf right?
<copumpkin>
they're enabled
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 2 new commits to master: https://git.io/vyIh9
<globin>
copumpkin: yeah that is a bugfix so that should get in anyway as long as it doesn't break other stuff
<pierron>
bennofs: They should evaluate to the same thing, and Nix used to be maximaly lazy, which means that you could cache an evaluation, and get the same result if you are evaluating it again.
<copumpkin>
globin: cool, thanks
eacameron has joined #nixos
<pierron>
bennofs: otherwise, the asDerivation view would have to carry some sort of caching mechanism.
<pierron>
bennofs: which would be similar to what I done in the shipping-security-update branch.
<bennofs>
pierron: but caching it is quite expensive, as you would have to also store the whole environment to make it safe
d762b440__ has joined #nixos
<pierron>
bennofs: less expensive than what we do today.
filterfish_ has quit [(Read error: Connection reset by peer)]
<pierron>
bennofs: I would appreciate if you can leave this remark on the pull request.
<bennofs>
pierron: there's a PR?
<pierron>
bennofs: yes, to zimbatm rfc repository
<pierron>
bennofs: or on the commit.
<bennofs>
pierron: ok, will do
<pierron>
bennofs: thanks.
estewei has joined #nixos
johnsonav has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] ttuegel opened pull request #23239: Organize KDE packages correctly (master...shuffle-kde) https://git.io/vyLe4
eacameron has quit [(Ping timeout: 260 seconds)]
<gchristensen>
copumpkin: can you do me a favor on your big machine?
<copumpkin>
"my"
<copumpkin>
:)
<copumpkin>
sure
eacameron has joined #nixos
<gchristensen>
can you run `nox-review --keep-going pr 23239 2>&1 | tee build.log`
<gchristensen>
in a screen session, it'll take a while
<NixOS_GitHub>
[nixpkgs] globin pushed 3 new commits to master: https://git.io/vyLeK
<NixOS_GitHub>
nixpkgs/master 65a2e30 Robin Gloster: varnish: fix i686 build
<NixOS_GitHub>
nixpkgs/master 4df663a Robin Gloster: verilator: 3.884 -> 3.900
<NixOS_GitHub>
nixpkgs/master b55c259 Robin Gloster: vfdecrypt: fix build
<copumpkin>
gchristensen: it's a-workin
<gchristensen>
great, thank you :)
<copumpkin>
failed quickly with Automatic merge failed; fix conflicts and then commit the result.
<gchristensen>
copumpkin: would you mind granting ttuegel access to the box for a bit?
<copumpkin>
not at all
<copumpkin>
his github key should have access now
<gchristensen>
great, thank you :)
<gchristensen>
oh ... what is the IP? hehe
ttuegel has joined #nixos
<copumpkin>
PM'd directly to ttuegel :)
<peti>
domenkozar: WFT!!!! Why are you messing with the r-updates jobset on Hydra.nixos.org without talking to me first???
<NixOS_GitHub>
[nixpkgs] jyp opened pull request #23240: tensorflow: depend on cudatoolkit 8 and cudnn 5.1 (master...tf) https://git.io/vyLvy
<peti>
domenkozar: The release candidate for R 3.3.3 is out and I was testing that new version on the r-updates branch. It sucks to have other people interfere with that effort. :-(
<domenkozar>
peti: sorry no harm done, just revert my merge
wizeman has joined #nixos
<schoppenhauer>
hi. did anyone manage to install gajim with omemo-support?
<peti>
domenkozar: I'm not talking about the merge. I'm talking about the fact that you've reconfigured the Hydra job to build another person's repo while I was using it.
mudri has joined #nixos
<domenkozar>
peti: it failed to build/evaluate
<domenkozar>
so I thought it's not used atm
<peti>
domenkozar: Well, you thought wrong, and a quick glance at the git history would have shown that the last commit was 12 hours old or so.
<peti>
Grrr.,
<peti>
domenkozar: I really wish you would have talked to me first.
<domenkozar>
peti: I don't understand the issue, you can do s/bla/peti/ and it's back
<schoppenhauer>
because it wants python-axolotl, and it has a setting extraPythonPackages, which I set to pkgs.python27Packages.python-axolotl
<schoppenhauer>
but still, it complains about deprecation
<schoppenhauer>
but apparently doesnt even find it.
<domenkozar>
peti: I agree I should have waited for your response first, but I wanted to get R updates ready for branch-off
<domenkozar>
just wanted to help
<domenkozar>
peti: my understand is that my work can be reverted in matter of seconds
<domenkozar>
is that not right?
<copumpkin>
domenkozar: can hercules have user-scoped builds? :)
<copumpkin>
:D
<peti>
domenkozar: Yes.
<peti>
I already have.
<domenkozar>
copumpkin: not (yet)
<domenkozar>
copumpkin: do you mean isolated from the rest of the builds?
<domenkozar>
peti: OK, sorry again
<copumpkin>
domenkozar: I just mean namespacing jobsets sort of how github does it. So perhaps the common builds would be under NixOS namespace (like on github), but I could create jobsets under copumpkin/ that would be mine
<gchristensen>
I suppose it sounds like it was a tricky situation to debug and understand what had happened
<domenkozar>
copumpkin: that's the plan
<peti>
domenkozar: Never mind. No harm done.
<domenkozar>
peti: that's the reason why I'm working on hercules-ci
<domenkozar>
since this global-everything-hydra is hard given the amount of people we have
freusque has quit [(Ping timeout: 260 seconds)]
<domenkozar>
but I should've really just cloned the jobset
<peti>
domenkozar: Personally, I don't like that R update because the new version fails its test suite on Nix. I have no clue why mention-bot didn't tag me on this PR.
<domenkozar>
peti: yeah, I figured we should improve that in future
<Kendos-Kenlen>
Hi :)
<bennofs>
copumpkin: i don't think I understand the issue regarding memory consumption of the current callPackage approach. We have an attrset for each package that contains the name of a dep and a pointer to the derivation of the dep. With the new approach, we still have that (buildDeps) but instead of a pointer to the deriv its a pointer to the package description (which can be viewed as a deriv with asDerivation)
<copumpkin>
bennofs: sorry, what's context?
globin_[m] has quit [(Ping timeout: 240 seconds)]
<bennofs>
copumpkin: oh sorry, I meant to address this to pierron ^^
<peti>
domenkozar: WTF are you doing? You are merging all kinds of controversial changes all over the place! I didn't want #22613 to go in yet, and I didn't want #23235 merged either. This is really annyoing. Why are you messing with the Haskell package set and the generic builder without talking to me first?
<Kendos-Kenlen>
What is the best way to use another JRE ? Create a package "jetbrains-jre" and depend on it or relink everything correctly in the folder ? (because I can't run executables from it now)
<domenkozar>
peti: remember, I'm a maintainer too. I reviewed those changes and saw no issues. But it's not set in stone, we can improve them
timor has joined #nixos
<domenkozar>
peti: afaik you can still review them?
<peti>
domenkozar: Well, then you can fucking maintain it when trouble arises. Don't bother tagging me with any issues that might come up.
<pierron>
bennofs: the memory consumption comes from the fact that callPackage & mkDerivation hold on memory in order to make them overridable. With the SOS, this is no longer needed, and once you have a derivation you only have a hash, or a set of hashes.
peti has left #nixos ["WeeChat 1.7"]
<gchristensen>
:(
<domenkozar>
there goes my goodwill for the day :D
* gchristensen
sends hugs to domenkozar
<bennofs>
that was weird, tbh
* gchristensen
also sends hugs to peti :(
ThatDocsLady_afk is now known as ThatDocsLady
<Kendos-Kenlen>
:|
timor has quit [(Ping timeout: 255 seconds)]
joelmo has quit [(Quit: Connection closed for inactivity)]
<zetok>
is kde5 supposed to work on 16.09 ?
<domenkozar>
dunno, we never had gatekeepers for nixpkgs
<gchristensen>
it is, zetok
<zetok>
well, it doesn't
<Kendos-Kenlen>
zetok: what's the problem ? It works well here
<gchristensen>
tired of it taking up your couch?
<gchristensen>
works well here, too
<zetok>
dunno, it failed to install with some error
<Kendos-Kenlen>
what's the error?
xadi has quit [(Quit: Leaving.)]
<zetok>
sec, I'll boot the vm and try again
<gchristensen>
ok, well I think we'd love to help you if you can help us understand what you're doing, how you've tried to do it, and what the expected results are, and what the actual results are.
<LnL>
pierron: do you think the example in the writing modules section should include disabledModules? wasn't sure if it really belongs there
<Kendos-Kenlen>
gchristensen: +1
<zetok>
I've enabled `services.xserver.desktopManager.kde5.enable` in configuration.nix, disabled kde4, and I've ran `nixos-rebuild boot`
<pierron>
LnL: I don't know, do what makes sense.
<pierron>
LnL: sorry for the non-response, I am working on something else at the moment, I will try to review these changes around midnight.
<copumpkin>
pierron: oh a while back I asked for "overridable modules" and you asked for details which I never gave. I guess I just mean that it's frustrating if an upstream module doesn't do what you want, because there's no easy way to override it. It always gets included unconditionally by module-list.nix, and sometimes it just sets the wrong flag on something and isn't configurable. To override that, I need to clone nixpkgs, make the change,
<copumpkin>
then repoint my entire system to my clone. It'd be great if I could somehow change e.g., the behavior of one tiny flag
<pierron>
copumpkin: LnL and I worked on it during the FOSDEM, and LnL finalized it in a PR that should hopefully be merged soon.
MercurialAlchemi has quit [(Ping timeout: 268 seconds)]
najib has quit [(Ping timeout: 240 seconds)]
puffnfresh has joined #nixos
freusque has quit [(Quit: WeeChat 1.4)]
rmeis_ has quit [(Ping timeout: 240 seconds)]
freusque has joined #nixos
johnsonav has joined #nixos
fre has quit [(Ping timeout: 240 seconds)]
nitrix_ is now known as nitrix
griff_ has joined #nixos
Rotaerk has quit [(Read error: Connection reset by peer)]
erasmas has joined #nixos
systemfault has joined #nixos
systemfault has quit [(Client Quit)]
fre has joined #nixos
freusque has quit [(Ping timeout: 240 seconds)]
Fare has joined #nixos
johann__ has joined #nixos
<goibhniu>
zetok: does it help if you add --fallback?
fre has quit [(Quit: WeeChat 1.4)]
sellout- has quit [(Quit: Leaving.)]
freusque has joined #nixos
<zetok>
goibhniu: nope
mojjo has joined #nixos
<goibhniu>
:/
johann__ has quit [(Client Quit)]
bin7me has joined #nixos
stepcut has joined #nixos
<mojjo>
hi! is anybody experienced with deploying with nixops to a vserver? I tested nixops just on a local virtual machine. On the vserver, I'd install nixos first I guess.. and then, how could I deploy to that machine?
<stepcut>
mojjo: pretty sure you just install NixOS on the vserver and then run nixops deploy to deploy to it. For your physical machine .nix file specific, deployment.targetEnv = "none";
<stepcut>
you might have to specific some extra kernel modules. For example, on Xen I have to do, boot.initrd.availableKernelModules = [ "xen_blkfront" ];
rmeis has joined #nixos
zraexy has joined #nixos
<stepcut>
in other news, What if there was a `cabal-install-2-nix` command that would use the cabal install dependency solver to pick packages hackage and then write out a default.nix or shell.nix ? Conceptually, that should work, right?
<NixOS_GitHub>
[nixpkgs] lsix opened pull request #23242: pythonPackages.libais: init at 0.16 (master...add_python_libais) https://git.io/vyLZO
<ekleog>
hmm, anyone knows of a way to allot more memory to a VM in nixos-rebuild build-vm ?
johann__ has joined #nixos
johann__ has quit [(Client Quit)]
johann__ has joined #nixos
<bennofs>
ekleog: virtualisation.memorySize (in MB)
<ekleog>
great, thanks!
reinzelmann has quit [(Quit: Leaving)]
cfricke has quit [(Quit: WeeChat 1.7)]
<mojjo>
stepcut: alright, I'll see after nix is installed over there. but one thing in advance: I guess I'd later run 'nixops deploy' from my local machine. Where would I specify the IP of the remote server and the credentials?
freusque has quit [(Quit: WeeChat 1.4)]
<stepcut>
mojjo: in your physical machine .nix file, you'd have something like, deployment.targetHost = "198.58.103.211";
<stepcut>
also make sure to have, services.openssh.enable = true;
griff_ has quit [(Quit: griff_)]
<mojjo>
alright, now I have an idea of it!
<copumpkin>
niksnut: any thoughts on sticking the userdata reconfig thing into initramfs?
<copumpkin>
that actually makes more sense to me, and would avoid the FOUC
<niksnut>
link?
<niksnut>
don't know what you're talking about :-)
<copumpkin>
niksnut: sorry, I meant the reconfigure-on-EC2-startup in initramfs. Don't have a link :) just responding to your thing about system-update.target and/or doing it after multi-user.target
ryanartecona has joined #nixos
<copumpkin>
the folks in #systemd are suggesting that system-update.target might not be right for this
<copumpkin>
and suggested initramfs
<copumpkin>
so on EC2, we'd enable networking in initramfs, reconfigure system, then fire up systemd
<copumpkin>
seems fairly straightforward in principle, but I'm not sure if there's stuff I'm missing
<stepcut>
mojjo: not sure about the credentials -- perhaps if you can do `ssh root@server` with out a password (aka, via ssh keys), then nixops just piggybacks on that
<niksnut>
copumpkin: not gonna happen, we'd need to have all of nixos-rebuild in the initrd right?
<copumpkin>
niksnut: yeah...
<copumpkin>
does it have a large closure?
<niksnut>
~100 MB
<copumpkin>
hmm
freusque has joined #nixos
<bennofs>
copumpkin: could you have a 3-stage init, with a second initramfs like env for the rebuilding?
<copumpkin>
I dunno, I guess we could try the different startup target
<copumpkin>
default.target symlinks to amazon-init.target, which depends on networking and not much else, reconfigures, changes default.target to the usual, and then either reboots or somehow starts up the other target
<bennofs>
copumpkin: you wouldn't even need to reboot
<stepcut>
mojjo: obviously linode (with KVM) is different from vserver -- but in both cases you are attempting to use nixops on a platform that is not natively supported, so there is crossover
<bennofs>
copumpkin: just systemctl isolate usual-target afterwards
<copumpkin>
okay, so I guess that's the sensible thing to do here
<copumpkin>
sigh, work :)
d762b440__ has quit [(Read error: Connection reset by peer)]
<copumpkin>
niksnut: does that approach make sense to you? I can try to hack it together later (I really want 17.03 to have a working EC2 image :P)
nschoe has quit [(Ping timeout: 240 seconds)]
<zetok>
goibhniu: actually nvm, it didn't fail yet – I saw curl fail and I thought that it would error out again, but apparently it's still working and is rebuilding stuff (ffmpeg now for some reason?)
<copumpkin>
is there a way to set -Q by default in nix.conf?
freusque has quit [(Ping timeout: 260 seconds)]
<goibhniu>
zetok: ah, good!
<niksnut>
copumpkin: I don't have a clear understanding of the problem
<copumpkin>
niksnut: the current switch-to-configuration.pl logic decides which targets to start based on which are already running, so when I call `nixos-rebuild switch` inside the service at bootup (before multi-user.target is active), switch-to-configuration doesn't try to start any new services that the userdata config requests
<copumpkin>
niksnut: I think it kinda worked by accident before :)
<copumpkin>
in that the reconfiguration took long enough for multi-user.target to have started up
<copumpkin>
so then it works
<niksnut>
right, that's why I suggested ordering it after multi-user.target
<copumpkin>
yeah
<copumpkin>
and my response to that is that it feels kind of awkward to do it that way because then you get a glimpse at the unconfigured system before it starts up
<bennofs>
niksnut: but then you see the unconfigured system for a short amount of time
<copumpkin>
that's what I've been calling FOUC
<copumpkin>
like the old websites that would load javascript/css after page loads and change appearance :)
<niksnut>
why is that a problem?
<niksnut>
I mean, the base AMI only has SSH
<niksnut>
sshd
<copumpkin>
it complicates other systems I need to hook up to it, which expect to be able to talk to the machine in the configured state
<copumpkin>
it means I need to add an additional set of wait conditions outside this
<copumpkin>
it also feels like a break in the model
<copumpkin>
I mentioned in the ticket, but ideally I'd have an EC2 API saying "give me a NixOS config and I'll give you an instance configured that way"
<copumpkin>
that's almost what we have, except it's "and I'll give you an instance that will eventually be configured that way, but probably won't be when you first see it"
takle has quit [(Remote host closed the connection)]
<niksnut>
sure, but booting a new machine is never an atomic operation
<copumpkin>
sure, but we do have the ability to make it seem like one
<bennofs>
copumpkin: sshd could still come up before the other services
<copumpkin>
not if we use the separate target
<bennofs>
copumpkin: well there's still a small amount of time in which one service might be up and the other hasn't finished starting yet
<niksnut>
it all depends on how you wait for the instance to be ready
<copumpkin>
and I get the usual dependency powers of systemd inside the config
pi3r has quit [(Ping timeout: 255 seconds)]
<copumpkin>
like if I have my custom service, I can say it should happen before sshd
<niksnut>
if "machine reachable via ssh" is the criterion, then sure, this breaks that
takle has joined #nixos
<copumpkin>
I'm not saying I can't work around it
<copumpkin>
I'm just saying it doesn't seem ideal, if we can do it better
takle has quit [(Remote host closed the connection)]
<copumpkin>
the config that the image was built with should be largely irrelevant to the end user
<niksnut>
I mean, if the only problem is to prevent ssh from starting, that's probably doable
<gchristensen>
that seems like a nightmare situation for debugging
<niksnut>
you can probably symlink /run/systemd/system/sshd.service to /dev/null
<lewo>
people that uses cloudinit also assumes the machine is not ready at startup.
<niksnut>
from the initrd
<copumpkin>
gchristensen: yeah, ideally I'd like it to boot into a rescue mode with SSH if it fails
<copumpkin>
niksnut: sure
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vyLBv
<NixOS_GitHub>
nixpkgs/master bccac38 Franz Pletz: microcode updates: prepend first in initrd...
<niksnut>
iirc, there also is some PAM mechanism to prevent logins
<gchristensen>
I don't see the need for such perfection, copumpkin. it isn't a "twas born perfect" scenario when you're using user-data, that only works if you precreate the AMI
<lewo>
copumpkin, what about generating an new image with your required configuration? I don't know aws, but the only problem I see with this on openstack is the image upload time
<copumpkin>
yeah, I do that all the time
<copumpkin>
but the boot-from-userdata is also pretty useful
<copumpkin>
niksnut: would I need to make a new default target to run something after multi-user.target?
<gchristensen>
yeah but it is definitely a two-stage deployment
<copumpkin>
gchristensen: it does't really have to be, fundamentally
johnsonav has joined #nixos
<niksnut>
copumpkin: no, you just order your service after multi-user.target
<gchristensen>
only if you pretend it isn't though
johann__ has quit [(Quit: Leaving.)]
<copumpkin>
niksnut: hmm, doesn't something need to want it for it to get installed properly?
<lewo>
copumpkin, if I understand well, your main problem is to know when the vm is "really" ready?
<niksnut>
I think you can make it wanted by multi-user.target
<bennofs>
copumpkin: you can still say that multi-user.target wants amazon-init, but order amazon-init after multi-user.target
<copumpkin>
oh
<gchristensen>
copumpkin: is this a change you're angling for 17.03?
<niksnut>
and magically, this will not be considered a circular dependency
<copumpkin>
gchristensen: yes, the current image is b0rked
<copumpkin>
niksnut: oh because ordering != dependency, right
<gchristensen>
borked because it is 2-stage, or borked because it is really borked?
<niksnut>
yes
<copumpkin>
b0rked because it doesn't work at all right now
<copumpkin>
you can add services.httpd.enable = true to your config
<copumpkin>
and the machine will boot up with no httpd running
<copumpkin>
if you log in and nixos-rebuild switch, it'll fire it up
<copumpkin>
not really ideal
<adev>
Hi Guys
<niksnut>
another workaround: use a timer unit that triggers N seconds after startup (not ideal I realize)
<copumpkin>
lewo: not quite
<gchristensen>
true enough, I'm hesitant though to slip in what is sounding like a non-trivial change
<copumpkin>
gchristensen: well, it doesn't work
<copumpkin>
it used to work :P
<gchristensen>
you broke it :P
<copumpkin>
yes, which is why I'm trying to fix it
<gchristensen>
yeah, fair enough
<bennofs>
isn't ordering it after multi-user.target pretty "trivial"?
<copumpkin>
yeah, I'll just do that for now
<adev>
just a little announcement : The Blue Brain Project recruit and is looking for someone with packaging, NixOS / nix packaging knowledges and with a software engineering background. If any of you might be interested : http://emploi.epfl.ch/page-142376-en.html
<gchristensen>
adev: :O
<adev>
The job is in Switzerland, Geneva, Swiss salary and Work visa granted for EU citizen if recruited
<pierron>
bennofs: I just saw your comment, I would not be able to reply soonish, but I wanted to say Thanks for the gist.
<niksnut>
adev: cool, you should send that to nix-dev as well
endformationage has joined #nixos
<ikwildrpepper>
garbas: see adev link for next nixos weekly
<gchristensen>
adev: BBP is around #200 in the TOPS500 right?
<musicmatze>
has someone a way to build a RAW image for nixos?
<musicmatze>
I have a cloud (openstack) account at my university, but botting from ISO does not work ... I guess I need a RAW image...
<musicmatze>
is there a way to build one?
<layus__>
nbp
stepcut has quit [(Ping timeout: 268 seconds)]
<bennofs>
samae: cabal2nix --shell generates a derivation that switches to a different version when using nix-shell compared to nix-build
<bennofs>
samae: look at default.nix, and note the "lib.inNixShell" call
<samae>
Oh !
layus__ has quit [(Quit: Leaving)]
layus has joined #nixos
<fpletz>
musicmatze: just a regular generic nixos image or one with a special config?
Kendos-Kenlen has quit [(Quit: Konversation terminated!)]
<NickHu>
I'm trying to update the qutebrowser package, but it requires updating dependencies - should that go in different commits/pull requests?
stepcut has joined #nixos
layus has quit [(Client Quit)]
<musicmatze>
fpletz: well, it must be bootable in openstack, from there I guess I have access to a TTY and I can configure it from there - if there is a way to include a custom config, that'd be really nice, though!
<musicmatze>
but for the first step I guess I don't need that! :-)
<samae>
bennofs: is there an explanation on why this is done?
<samae>
Removing this check fixed my problem (maybe not in the right way)
<musicmatze>
NickHu: In different commits, but in the same PR - at least that's what I do all the time.
<copumpkin>
niksnut, bennofs: after multi-user.target seems to work fine, thanks. Can't say I love it working this way, but at least it isn't horribly broken :)
<bennofs>
samae: it is so that you can, for example, run ghci inside the nix-shell and get a ghci that lets you import the dependencies
<fpletz>
musicmatze: unfortuantely the file is not very organized, but you should get the gist
<fpletz>
musicmatze: there's also code for custom iso images and netboot images
<NixOS_GitHub>
[nixpkgs] copumpkin closed pull request #23224: switch-to-configuration: don't try to figure out active targets ourselves (master...switch-to-configuration-fixes) https://git.io/vyk9k
<musicmatze>
awesome, i will start digging! Thanks fpletz
<bennofs>
samae: in general, the idea behind that is that you're unlikely to build the package the same way as Nix would do it (nix does cabal configure ...pass explict paths to each dependency here...)
<bennofs>
samae: but if you use nix-shell for development of haskell packages, you probably don't want to have to pass the paths to each dep explicitly
<samae>
my use case here is that my haskell package is a dependency of another nix-shell (which pulls other stuff)
<bennofs>
samae: in general, the nix-shell --shell expression is meant primarly for usage with nix-shell. That you can nix-build it as well is just added convenience
<samae>
ok
dustmote has quit [(Quit: bye)]
takle has joined #nixos
<samae>
cabal2nix output something incomplete (missing the haskellPackages.callPackage for instance)
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vyLgU
<NixOS_GitHub>
nixpkgs/master 6018cf4 Dan Peebles: amazon-init.service: fix starting services at startup...
<NickHu>
musicmatze: Should I replace the maintainer with me, or add myself to the list of maintainers? Does the order of the list matter?
stepcut has quit [(Remote host closed the connection)]
arjen-jonathan has quit [(Ping timeout: 255 seconds)]
<musicmatze>
NickHu: The order does not matter. If you want to maintain these packages (you don't have to), feel free to add yourself. Never remove another maintainer, though.
stepcut has joined #nixos
dustmote has joined #nixos
stepcut has quit [(Read error: Connection reset by peer)]
filterfish_ has quit [(Remote host closed the connection)]
filterfish_ has joined #nixos
stepcut has joined #nixos
magnetophon has quit [(Remote host closed the connection)]
<endformationage>
What does the branch-off for 17.03 mean for existing packages that need updates, but aren't in master yet? Can they still get updated on the release branch?
dustmote has quit [(Client Quit)]
teknico has left #nixos []
<NickHu>
Should I just set nix.useSandbox to be true? Why is it false by default?
__Sander__ has quit [(Quit: Konversation terminated!)]
<copumpkin>
it's a bit slower
<copumpkin>
many of us think it should be true by default anyway :)
jgertm has quit [(Ping timeout: 260 seconds)]
<NickHu>
I only ever build packages when I am going to submit a pr for it so I guess I should just set it to true
filterfish_ has quit [(Read error: Connection reset by peer)]
<musicmatze>
fpletz: Maybe you have another hint for me: I'm trying to build a nixos image on a debian machine, via a commandline call from the file mentioned in this commandline, which is also the commandline I executed to build: 'NIXOS_CONFIG=`pwd`/nixos/modules/virtualisation/nova-image.nix nix-build '<nixpkgs/nixos>' -A config.system.build.novaImage'
<musicmatze>
fpletz: although, it tells me: Could not access KVM kernel module: Permission denied - and: failed to initialize KVM: Permission denied
<musicmatze>
the debian machine is a virtual machine in the mentioned openstack infrastructure... :-/
<copumpkin>
niksnut: any thoughts on my comments about impure derivations?
<fpletz>
musicmatze: hrm, not sure if you can disable kvm here, let me have a look
<jophish_>
What's necessary to get opengl applications working on non-nixos?
adev has quit [(Quit: Ex-Chat)]
<jophish_>
I've symlinked my libGL.so to /run/opengl-driver/lib, however glxgears is still looking for swrast
<jophish_>
in fact, LD_DEBUG tells me that that's not even being looked for
<fpletz>
musicmatze: unfortunately it looks like kvm is hardcoded somewhere, at least it's not easily configurable
<jophish_>
and the libGL in mesa-noglu is being used
<fpletz>
I have not yet found the place, but it call runInLinuxVM from pkgs/build-support/vm/default.nix
kthnnlg has joined #nixos
<kthnnlg>
Anyone have a quick tip on the best method to install nodejs 6.10?
<doublehp>
i don't have any sound when I play youtube in firefox 51. other apps work fine using alsa and pulse (at the same time, it works for me). there are bugs about FF lacking sound configuration interface. How do I dig ?
<doublehp>
I have reinstalled Firefox with some options in .nixpkgs/config.nix , it reinstalled more flash and java, but did not help
<Arguggi>
(I'm still very inexperienced with nix) I'm writing a haskell program that needs postgres and also needs to create a user/database and import an sql file. What would you say is the best way to test/set this up?
kthnnlg has quit [(Ping timeout: 240 seconds)]
bfrog has joined #nixos
<auntie>
Arguggi: in nixos, if you're doing any system stuff (such as creating users or starting a daemon) you want to write a nixos "module"
<auntie>
Arguggi: those are located in nixpkgs/nixos/modules
<auntie>
Arguggi: so.... you would ideally end up writing a nix "derivation" under nixpkgs/pkgs, and then once you have your haskell program compiling you would write a nifty "module" under nixpkgs/nixos/modules to do the system-level stuff
<auntie>
lots of nixos services will create a user/database thing... so there should be lots of examples in the nixpkgs repo
griff_ has joined #nixos
<Arguggi>
auntie, ok thanks. Let me check I'm getting this right. I compile my haskell program with nix as usual, then I write a module which takes care of setting up postgres, config files ecc
<auntie>
yup
<auntie>
the module you can put in nixpkgs... if it's site-specific, you can also import it in your configuration.nix
<auntie>
technically speaking, configuration.nix /is/ a nixos module
<auntie>
(iinm)
<simpson>
YNM.
<Arguggi>
any idea of any modules that interact with posgres so I can get an inkling of what I should do?
<cransom>
the postgres module.
<LnL>
^ services.postgresql
DutchWolfie has quit [(Quit: Konversation terminated!)]
<cransom>
http://nixos.org/nixos/options.html is a wealth of information (or also, man configuration.nix and search around for things of interest)
<Arguggi>
LnL, thansk
<auntie>
when you write your module, you would probably want to expose the database name as a parameter
<LnL>
yes, also check out man configuration.nix and the nixos-option command
ronny has quit [(Ping timeout: 255 seconds)]
<auntie>
hrm... I ended up using linux_4_9.dev to get to the kernel sources
<auntie>
seems like a poor solution :/
tokudan has joined #nixos
<Arguggi>
I'll ask something else since you all seem so helpful. How far behind are the pkgs usually? For example the kernel, since I'll probably need 4.10+ ?
<auntie>
the nixos-unstable channel is the one typically used by developers...
takle has joined #nixos
<auntie>
that channel is held back until the nixos test suite passes
<auntie>
sometimes it gets held back by a month or two, but usually it's only a week behind
ronny has joined #nixos
<auntie>
otherwise, the stable channel is, what, twice a year? like ubuntu
<LnL>
yep, we are branching off today :)
takle_ has joined #nixos
<LnL>
auntie: you want the currently active kernel?
<Arguggi>
ah ok I can live with ~1 month, I'm on arch so I'm pretty used to it. Thanks auntie
<auntie>
LnL: yes? I don't know if that's possible though...
takle has quit [(Read error: Connection reset by peer)]
<auntie>
unless there is an even more convoluted nix-shell command ;P
<auntie>
I'm just uh, trying to set up a kernel module development environment
filterfish_ has quit [(Read error: Connection reset by peer)]
hotfuzz has quit [(Read error: Connection reset by peer)]
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vyLy6
<NixOS_GitHub>
nixpkgs/master 8def08a Dan Peebles: apache-kafka.service: pass in log4j config more explicitly...
mudri has quit [(Ping timeout: 260 seconds)]
<LnL>
auntie: if you're on nixos you can use 'with import <nixpkgs/nixos> {}; config.boot.kernelPackages.kernel.dev'
hotfuzz has joined #nixos
<icetan>
has anyone had problems with cron not running on nixos?
<copumpkin>
icetan: I thought it was disabled by default
<gchristensen>
you'd need to set services.cron.enable to true
<LnL>
I think most things on nixos use systemd timers
katyucha is now known as Guest19083
Guest19083 has quit [(Killed (adams.freenode.net (Nickname regained by services)))]
katyucha_ is now known as katyucha
<gchristensen>
globin: hows it going?
katyucha_ has joined #nixos
<icetan>
copumpkin: yes, but the docs say it gets enabled when you set services.cron.systemCronJobs
<icetan>
sorry, user crontabs run
<icetan>
it's the system crontab not running
<mojjo>
I'm installing nixos on a vserver. the 'nixos-install' process worked fine, until I'm getting 'error, will not proceed with blocklists; installation of grub failed'... any suggestions?
<globin>
gchristensen: just finishing my meal then I'll branch off and will prepare the zhf ticket
<gchristensen>
yayayay
<NixOS_GitHub>
[nixpkgs] Ma27 opened pull request #23247: nodePackages.yo: init at 1.8.5 (master...init/nodePackages.yo) https://git.io/vyLSR
<auntie>
LnL: awesome. that was exactly what I needed
Bodil has quit [()]
Bodil has joined #nixos
kthnnlg has joined #nixos
<icetan>
LnL: maybe I should use a systemd timer instead of a system cron job?
hamishmack has quit [(Ping timeout: 240 seconds)]
fre has joined #nixos
justan0theruser has quit [(Ping timeout: 268 seconds)]
<zetok>
Arguggi: 4.9 seems to be available on 16.09
<zetok>
Arguggi: aside from kernel at least some packages are quite outdated
<Arguggi>
zetok, even on the unstable channel?
<zetok>
if unstable = master, yes
stepcut has quit [(Remote host closed the connection)]
stepcut has joined #nixos
stepcut has quit [(Ping timeout: 260 seconds)]
<hyper_ch>
hi, I can't get VirtualBox interface to start anymore... it keeps telling me: Effective UID is not root (euid=1000 egid=100 uid=1000 gid=100) (rc=-10) Please try reinstalling VirtualBox. but existing vms I can start fine with vboxmanage startvm
justan0theruser has joined #nixos
fre has quit [(Ping timeout: 240 seconds)]
johann__ has quit [(Quit: Leaving.)]
ryanartecona has quit [(Quit: ryanartecona)]
<mojjo>
I'm creating partitions for a nixos setup on a vserver. I'm not familiar with this kind of things. Is it right, that I need at least two? a boot and one for the rest?
<mojjo>
and if so, of which types shoud they be?
Sonarpulse has joined #nixos
Sonarpulse has quit [(Remote host closed the connection)]
<copumpkin>
mojjo: nah, a single partition works fine
Sonarpulse has joined #nixos
<copumpkin>
most of our images are a single ext4 partition
<NixOS_GitHub>
[nixpkgs] bjornfor opened pull request #23248: eclipse: add aliases for "latest" versions (master...eclipse-aliases) https://git.io/vyLFl
<icetan>
LnL: I upgraded to unstable/nixos-17.03pre101839.53a2baa just now
<mojjo>
copumpkin: I did so in the beginner, but I got an error in the install process..
<mojjo>
at 19:38
<mojjo>
(oh, that might be just my timezone, well, some lines above...)
<LnL>
icetan: have you tried to reboot? there are some issues related to the security.wrappers change when upgrading in place at the moment
<icetan>
aha, will try. Cheers!
<NixOS_GitHub>
[nixpkgs] globin created release-17.03 (+1 new commit): https://git.io/vyLFF
<NixOS_GitHub>
nixpkgs/release-17.03 6c9fb36 Robin Gloster: release 17.03-beta
<LnL>
^ WOOT!
<copumpkin>
zomg globin
<gchristensen>
WHAAAAT
<gchristensen>
!m globin
<[0__0]>
You're doing good work, globin!
<gchristensen>
now I have to update all my scripts :P
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to master: https://git.io/vyLby
<NixOS_GitHub>
nixpkgs/master b7d15ed Robin Gloster: bump version to 17.10...
<copumpkin>
Mic92: not sure what you're saying in your latest comment on the switch-to-configuration.pl PR
edvorg has joined #nixos
<icetan>
LnL: still nothing :(
<icetan>
woop woop, is 17.03 is stable?
<copumpkin>
it's been branched off
<copumpkin>
now begins stabilization :)
<LnL>
not yet, but soon
<icetan>
wow, great job
<Mic92>
copumpkin: forgot that for a moment. I think I will read more systemd code, but there are some interesting systemctl options, which might solve the problem.
<gchristensen>
mojjo: I'm not sure if you answered or not, did you run `systemctl start sshd`?
<kthnnlg>
Hi All, has anyone managed to install on nixos version >= 6.9 of nodejs? Thanks
<LnL>
kthnnlg: we have 7.6.0 on unstable
<mojjo>
gchristensen: oh, I thought this was done by the declaration in the config file. Anyways, yes, now I did so and at least the client connects!
<kthnnlg>
ok super, could you point me to the instructions on how to use unstable? if possible, i'd prefer to use unstable for just this package and a few others
<gchristensen>
mojjo: the CD is a bit special, no config change needed
<gchristensen>
just the systemctl start sshd
<mojjo>
.. it does not accept my password however, which is the default ""
<LnL>
kthnnlg: for you nixos configuration or some other expression?
AllanEspinosa has joined #nixos
<copumpkin>
pierron: if I have several modules I'm importing and I want to make something available to all of them in their module arguments, how do I do that again?
<kthnnlg>
LnL: I'm not sure I understand. I'd like if possible to install the package via configuration.nix. But if the prefered method is to use nix-env -iA ... then that's ok too
<copumpkin>
pierron: something like _args, right?
<copumpkin>
pierron: (this is not for nixpkgs use, just local config I'm refactoring)
<kthnnlg>
Could I do something like the following? { config, pkgs, ... }: let unstable = import <nixos-unstable> {}; in { environment.systemPackages = [ unstable.packagehere ]; }
<LnL>
^ yes that should work
<copumpkin>
_module.args, aha
<eacameron>
What's the easiest way to put a "writeScript "dosomething ..." derivation on my profile PATH?
<LnL>
copumpkin: yes that's how pkgs is added to the args
<kthnnlg>
LnL: thanks :) Now, if I add that unstable channel, will there be any side effects, such as having all my system packages upgrade?
<Mic92>
eacameron: use pkgs.writeScriptBin instead of writeScript
<LnL>
kthnnlg: if you use the packageOverrides approach it might, but if you just add unstable.foo to your systemPackages that will be the only thing from unstable
<eacameron>
Mic92: Ah interesting. In my case I want it in systemPackages
<kthnnlg>
LnL: thanks again :) is there an easy way to know if I
<mojjo>
;-) ssh does not work with empty passwords. if root has a pwd, it works. thx!
<Mic92>
eacameron: my `all` package works similar to systemPackages, both will work
<kthnnlg>
if I'm using package overrides
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vytU5
<eacameron>
Mic92: Ah so the trick is to use writeScriptBin?
<Mic92>
eacameron: yes, will create a bin directory in the packet and put the script there
<Mic92>
*package
<eacameron>
Mic92: I think that's what I was missing then.
freusque has quit [(Ping timeout: 260 seconds)]
<LnL>
kthnnlg: in my example everything that depends on docker will use the updated version, but I don't think there's an easy way to see what packages change because of it
ThatDocsLady has quit [(Quit: Arma-geddin-outta-here!)]
[0x4A6F] has quit [(Ping timeout: 240 seconds)]
griff_ has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 5 new commits to master: https://git.io/vytkK
<NixOS_GitHub>
nixpkgs/master 1aa6d77 Leon Isenberg: haskell: add semigroups dependency to Quickcheck for GHC < 8
<NixOS_GitHub>
nixpkgs/master 24c9361 Leon Isenberg: haskell: break dependency cycle between QuickCheck and semigroups for GHC < 8...
doublehp has quit [(Quit: Leaving.)]
<kthnnlg>
LnL: I think I can kinda see it. For example, your configuration.nix presupposes the command `nix-channel --add http://nixos.org/channels/nixos-unstable nixos-unstable` was issued. The binding `unstable = import <nixos-unstable> {};` mentions the unstable channel specifically. So, there should be something analogous binding somewhere for pkgs (which is my default). So, since all of my current packages in the list are of the
stepcut has joined #nixos
greymalkin has joined #nixos
reinzelmann has joined #nixos
reinzelmann has quit [(Remote host closed the connection)]
<philipp[m]>
Anybody else getting build errors for python2.7-libasyncns-python-0.7.1?
obadz has quit [(Ping timeout: 260 seconds)]
<Mic92>
philipp[m]: there was a fix pushed last night
<Mic92>
glibc breakage
<philipp[m]>
glibc breakage ist best breakage :-D
obadz has joined #nixos
<copumpkin>
if I builtins.readDir ./., that won't copy ./. into the store, right?
<auntie>
I'm looking for a better workflow while developing software with nixos... my problem is that updating the fetchgit hash is too tedious
<copumpkin>
builtins.fetchgit ;)
<copumpkin>
niksnut added that fairly recently, not sure if it's in any release yet
<viric>
Hello all - I need to use some service with non-geeks to share files/directories, so everyone can add/remove/move them. Any suggestions? The more FOSS clients accepted, the better.
<auntie>
right now my workflow is "make source change -> git add -> git commit --amend -> nix-prefetch-git -> copy/paste hash -> nix-build"
<copumpkin>
auntie: if it's purely local, just use a path reference
<viric>
I wish I could use sftp or some fuse-able thing
<copumpkin>
src = ./.
<auntie>
copumpkin: does that really work? :s
<copumpkin>
yeah
<copumpkin>
generally works best if your working dir isn't the path you're referring to
<copumpkin>
otherwise it'll get junked up by result links and such
<copumpkin>
auntie: when you refer to a path without quotes, nix copies that path into the store as a fixed-output derivaiton
<auntie>
copumpkin: I have been living a lie :|
<auntie>
thank you xD
<copumpkin>
so if the contents doesn't change, nix won't do any work
<auntie>
right
<copumpkin>
but keep in mind that you nix-build and it drops a result link into the current directory
<mojjo>
is there a way to set nixos-options imperatively? (without changing the config file)
<copumpkin>
that'll cause it to create a new derivation in the store because it has a new result link :)
Sonarpulse has quit [(Ping timeout: 260 seconds)]
<auntie>
haha, that's not good
<copumpkin>
that's what I meant about watching out for --no-out-link or making it point to a subfolder
<auntie>
oh well :P
<auntie>
I'll do that then
<copumpkin>
well, you could also use lib.cleanSource which does something like that
<copumpkin>
I think that copies to the store and filters out the crap
<auntie>
I wonder if it would filter vim files too
<eacameron>
Can anyone recommend an easy way to get notifications when systemd units or timers fail?
<greymalkin>
Quick nixops-sorta question: Is there a way to create a read/write directory as part of a deployment? Say, /var/www-user-data or some such?
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to release-17.03: https://git.io/vytYo
<NixOS_GitHub>
nixpkgs/release-17.03 69d75ce Robin Gloster: release-notes: add new services for 17.03 + misc additions
<maurer>
greymalkin: I do that by attaching an explicit separate volume
<auntie>
eacameron: umm... you mean like sendmail?
<maurer>
greymalkin: but that may or may not be ideal for your case
<eacameron>
auntie: That would be a viable option, yes.
<maurer>
(so that I have one volume for my system software, controlled and managed by nixops, and another volume which nixops mounts, which is just the intentionally mutable state)
<maurer>
(separating data which can be regenerated easily and is pretty static from data which it's important to back up etc)
<eacameron>
auntie: But I'm interested in the general case and not excited about manually plumbing sendmails everywhere
<auntie>
eacameron: last time I used sendmail was with upstart... but systemd should be able to run some command on failur
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vytOf
<arianvp2>
I have a nix expression for my haskell package
<maurer>
is an example of hydra engaging in this practice
<greymalkin>
maurer: K, that will work (even the directory already existing tells me that it worked :)
<arianvp2>
but every time i do nix-build I get a different hash
<arianvp2>
:/
<auntie>
(it's always a good idea to have a log server anyway)
<arianvp2>
so it never caches it!!
<eacameron>
auntie: I was just about to say that. I suppose that's the *correct* answer. I just have these cheapy wordpress servers and don't care about much other than total failure
<eacameron>
auntie: Total failure of certain components that is...system outage I can monitor easily.
<auntie>
eacameron: that is until the FBI comes knocking at your door asking what the heck your servers are doing ;P
<copumpkin>
arianvp2: are you using a local file reference and nix-build is dropping a result link into it each time?
<arianvp2>
yes src = ./.
<copumpkin>
:)
<arianvp2>
and results in result
<copumpkin>
so run nix-build --no-out-link
<arianvp2>
... wait
<arianvp2>
is it using results as a build input?
<arianvp2>
xD
<copumpkin>
yeah
reinzelmann has joined #nixos
<arianvp2>
lol
<copumpkin>
or I think lib.cleanSource will ensure it's not in there
<eacameron>
auntie: The FBI won't need to ask ;)
<copumpkin>
can't remember what it filters out
phant0mas has quit [(Ping timeout: 240 seconds)]
<arianvp2>
why doesnt cabal2nix do this automatically :/
<auntie>
it looks like cleanSource does filter out the results symlink
<arianvp2>
okay so how would that look like?
<auntie>
but... something is still dirtying my build
<auntie>
it just... makes a new derivation with all of the "crap" removed
<eacameron>
auntie: I think it's a bit more than that. Paths are not derivations.
<sphalerite>
Is there a way to get the nvidia legacy drivers on nixos?
<sphalerite>
(340.xx)
<gchristensen>
sphalerite: nvidia_x11_legacy340
griff_ has quit [(Quit: griff_)]
<eacameron>
arianvp2: If you want to add your own filter rules, I'd combine them with cleanSourceFilter like this: builtins.filterSource (name: type: lib.sources.cleanSourceFilter name type && myFilter name type)
<sphalerite>
gchristensen: so I put that in hardware.opengl.videoDrivers?
<arianvp2>
zgot it
<arianvp2>
thanks guys
<gchristensen>
not sure about opengl, but I put it in services.xserver.videoDrivers
<arianvp2>
I just bought a backup hard disk to backup my laptop
<arianvp2>
so I can install nixos on it :)
<arianvp2>
im sold
<arianvp2>
been running Nix on arch linux for the past month
<arianvp2>
and nixos on my server
<eacameron>
arianvp2: welcome to city 17; it's safer here. :P
rmeis has joined #nixos
<sphalerite>
gchristensen: yep that worked, thanks!
<LnL>
arianvp2: when I started using nix thought that ./. was just never cached, like --check :p
<gchristensen>
yep!
<arianvp2>
hehe
<arianvp2>
it seems like a stupid default though
<arianvp2>
should at least spit out a warning
<sphalerite>
gchristensen: actually no, it's using nouveau x)
<gchristensen>
you may need to reboot?
<sphalerite>
I did
<maurer>
sphalerite: I found that I needed to blacklist nouveau to make it not load
<maurer>
There may be a more proper way to stop it, but this works
<eacameron>
arianvp2: totally agree, cabal2nix should have a better default
<LnL>
arianvp2: well it does make sense, ./. is nothing special it just copies everything to the store including symlinks
<sphalerite>
maurer: cool, will try that. Thanks
<zetok>
why does `firefox` package install ff nightly instead of the stable release?
<sphalerite>
zetok: it's branded as nightly because of trademark issues with a modified version of the browser
<eacameron>
LnL: But the point is that you never ever want .stack-work (for example) to get into the store. If you need it, then you're the one who needs to override. But instead every has to override to get the obvious correct behavior in most cases.
<sphalerite>
zetok: it's actually stable
edvorg has quit [(Ping timeout: 240 seconds)]
stepcut has quit [(Read error: Connection reset by peer)]
<zetok>
>.<
<maurer>
zetok: tl;dr if you ship a binary (which the nixos cache does) you can't put the regular firefox branding on it
<maurer>
zetok: without a special deal with mozilla
<maurer>
Do we have a switch in the build that enables normal branding?
<copumpkin>
ಠ_ಠ
<sphalerite>
yes we do
<zetok>
and the iceweasel patches from debian can't be used?
<sphalerite>
but of course then it has to build on your own machine
<sphalerite>
they could
johnsonav has joined #nixos
<maurer>
zetok: Even debian ships firefox with branding now iirc? I think they came to an agreement
<zetok>
yeah
<maurer>
sphalerite: yeah, I just figured if we told zetok how to get that flag, it might fix the issue for him
<maurer>
even if he had to wait a bit on the build
<sphalerite>
can't remember it off the top of my head though :p
<zetok>
well, branding is not an issue, I was just surprised that seemingly nightly got installed when I wanted stable ff
<sphalerite>
enableOfficialBranding
<zetok>
hmm
<maurer>
Oh, uh, there's another option too
<zetok>
AFAIK it's possible to ship unmodified ff binaries with branding enabled
<maurer>
looks like we have a firefox-bin package
<maurer>
which uses a stock firefox
<zetok>
:D
<maurer>
rather than a nix built one
tokudan has quit [(Quit: Leaving)]
<maurer>
you could try that
<gchristensen>
but also, `firefox` _is_ the stable ff.
Havvy has quit [(Quit: Computer Restarted or Restarting IRC)]
<maurer>
gchristensen: this has already been explained to him
rmeis has quit [(Ping timeout: 264 seconds)]
jensens has quit [(Ping timeout: 260 seconds)]
<maurer>
he was just surprised by the branding
<gchristensen>
just saying it again to be clear.
<zetok>
`nix-env -qaP | grep firefox` doesn't show firefox-bin
<gchristensen>
zetok: it is considered unfree
<eacameron>
Anyone know of a good self-hosted OSS log aggregation tool?
<gchristensen>
eacameron: logstash is nice
<eacameron>
gchristensen: That's OSS and self hosted?
<arianvp2>
but... yeh I must say, stuff is a bit rusty at times
<arianvp2>
systemd project sees this stuff as side-projects. so expect bugs.
<arianvp2>
if you want something production ready. use logstash or the likes :P
<eacameron>
arianvp2: Either way this is useful material to look into
<arianvp2>
but it's totally possible with 'just' nixos and systemd
<arianvp2>
i'll be writing a blog post about this stuff real son
<arianvp2>
soon*
<arianvp2>
i'll ping it here
<eacameron>
arianvp2: Plz do
<zetok>
gchristensen: yeah, from what I see it would seem that libre patches are not applied, thus the kernel is also unfree :D
<gchristensen>
I don't know
<gchristensen>
if you're interested, please do research and file a ticket with your findings
<zetok>
btw, it's quite misleading to have packages named `linux-*` and store them in directory named `kernel`
<sphalerite>
Is there a neat way to have two different configurations available with a choice between them on the bootloader? Ideally both configurations would be built at once by nixos-rebuild. (Use case: allowing switching between video drivers without a rebuild)
ryanartecona has quit [(Quit: ryanartecona)]
ryanartecona has joined #nixos
<zetok>
gchristensen: sure, I'll open an issue
<zetok>
right now kernel is freely distributable, but not free
<gchristensen>
perhaps it is marked with the `unfreeRedistributableFirmware` license?
<zetok>
um, I don't know with what it's marked in nixos
<zetok>
and last time I've checked, package had an unnecessary dependency on opencv, on which qTox doesn't depend since at least a year
<zetok>
gchristensen: as soon as I'll figure out how things work :)
<LnL>
gchristensen: do you know of some way to show that in git log? that would be nice
<gchristensen>
LnL: the signatures?
<gchristensen>
LnL: log.showSignature = true :)
<LnL>
just if one exists
hiratara has joined #nixos
stepcut has joined #nixos
griff_ has joined #nixos
<Acou_Bass>
eey folks, is there any place i can see a log for what gets updated when i have the configuration option to auto-run the nixos-rebuild switch --upgrade (so i can, for example, wake up in the morning, view log and see what updated while i was in bed)
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vytBb
<NixOS_GitHub>
nixpkgs/master 81b43cc Vladimír Čunát: 17.09 release notes: fix typos
jgertm has quit [(Ping timeout: 260 seconds)]
<viric>
zetok: yes, very outdated
<viric>
zetok: if multiuser chats worked well enough...
<viric>
zetok: (Keeping conversations through shutdowns)
<NixOS_GitHub>
[nixpkgs] juliendehos opened pull request #23251: hieroglyph: init at 0.7.1 (master...hieroglyph) https://git.io/vyt0R
indi_ has joined #nixos
JagaJaga has quit [(Ping timeout: 240 seconds)]
stepcut has quit [(Remote host closed the connection)]
<cransom>
Acou_Bass: `journalctl -f -u nixos-upgrade.service --since "24 hours ago"` would have that info.
<Acou_Bass>
ahhh cool thanks cransom ;D
FRidh has quit [(Remote host closed the connection)]
<Acou_Bass>
i think my main reasoning is i dont know if/when i get a kernel or other such update that requires a boot
mkoenig has quit [(Remote host closed the connection)]
<slyfox>
OSError: [Errno 28] No space left on device
<dcz__>
yeap i know
<dcz__>
i thought it should replace the old one with new one
<dcz__>
at least..
<clever>
dcz__: systemd will refuse to let you mount the "wrong" filesystem to /boot, you need "nixos-rebuild test" to update fstab, after you fix fileSystems."/boot"
hamishmack has quit [(Read error: Connection reset by peer)]
<dcz__>
there is nothing wrong about filesytems, i didnt get you what u meant
<viric>
Acou_Bass: well, I know that "24 hours ago", but I have no idea what syntax brings to that
<viric>
that "24 hours ago" works
<Acou_Bass>
not sure what you mean by that
hamishmack has joined #nixos
<Acou_Bass>
its the --since flag that takes args like that (it can take them in a couple of other syntax's too)
<sphalerite>
How do I use nixops on physical machines with an existing nixos installation?
<clever>
dcz__: ah, last time i was fixing that, i had deleted and remade /boot bigger
<gchristensen>
sphalerite: you have to put a nixos installation on first
<sphalerite>
gchristensen: I have. "physical machines with an existing nixos installation"
<viric>
Acou_Bass: to me it looks complicated
<Acou_Bass>
i think it takes cron-like syntax too
<Acou_Bass>
if that appeals to you
<gchristensen>
sphalerite: oops, I thought "without". You need the "none" backend
Kingsquee has joined #nixos
<sphalerite>
doesn't seem to be in the manual, are there any docs on it?
DutchWolfie has quit [(Quit: Konversation terminated!)]
athan has joined #nixos
<sphalerite>
oh, I see. Thanks!
<gchristensen>
you're welcome!
<sphalerite>
and how do I give the nixops-running-machine access to the physical machines?
<clever>
sphalerite: when the targetEnv is set to none, it will just ssh into root, and nix-copy-closure to upload a pre-built nixos, but the config you give to nixops has to keep the machine bootable
<sphalerite>
I'm guessing I need to install the keys manually
<clever>
sphalerite: yeah
<sphalerite>
oh jeez it's almost 11pm and I'm still at uni
<sphalerite>
where do I find the keys to install them on the machines?
<clever>
i'm guessing the main key in ~/.ssh/ will work
<mojjo>
I'm installing (still ;-) nixos on a vserver. Or well, remotely the system is installed now. My first deployment attempt gives me two errors (http://pastebin.com/NeP7qvrf) . I'm not quite sure if I understood this correctly: Will nixops create a seperate system remotely and it needs another partition for this, or how will it handle this?
<pierron>
LnL: Can you squash the commits into one, or should I do it from the PR?
<LnL>
I'm here
<qknight>
pierron: did you get my last message?
Rizy has joined #nixos
griff_ has quit [(Quit: griff_)]
<LnL>
pierron: sure, just a single commit?
stepcut has quit [(Remote host closed the connection)]
athan has quit [(Ping timeout: 240 seconds)]
griff_ has joined #nixos
ryanartecona has quit [(Quit: ryanartecona)]
stepcut has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 5 new commits to release-17.03: https://git.io/vytVx
<NixOS_GitHub>
nixpkgs/release-17.03 13d29e7 Leon Isenberg: haskell: add semigroups dependency to Quickcheck for GHC < 8...
<eacameron>
I cannot figure out a good way to run scripts from systemd. I want to run a backup script from systemd timer but it fails when run through systemd. I'm using borg which needs ssh but of course the ENV is completely bare when systemd runs it. Is there an easy way to just run the script as a user with that users full env?
mojjo has quit [(Ping timeout: 255 seconds)]
indi_ has quit [(Remote host closed the connection)]
stepcut has quit [(Ping timeout: 268 seconds)]
indi_ has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz closed pull request #22154: phpfpm service: one service per pool for isolation (master...refactor/phpfpm-service-per-pool) https://git.io/vMppa
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to release-17.03: https://git.io/vytrk
<NixOS_GitHub>
nixpkgs/release-17.03 f0e9dd5 Franz Pletz: phpfpm service: one service per pool for isolation...
<NixOS_GitHub>
nixpkgs/release-17.03 3433c1a Franz Pletz: phpfpm service: add target and slice...
<pierron>
qknight: I don't think I did.
<pierron>
qknight: the private one?
<pierron>
LnL: yes, please, such that I can rebase it on master.
<pierron>
LnL: without loosing the authorship by squashing it
<globin>
gchristensen, domenkozar, niksnut, fpletz, ikwildrpepper: I'd personally like to put my gpg key somewhere on the website similar to the security team to have an additional source of trust for checking gpg signed tags etc.
reinzelmann has quit [(Quit: Leaving)]
<globin>
gchristensen, domenkozar, niksnut, fpletz, ikwildrpepper: any idea where it would fit? the security page doesn't really fit
griff_ has joined #nixos
takle_ has quit [(Remote host closed the connection)]
batdog has quit [(Remote host closed the connection)]
<sziszi>
hey, I tried to update my nixOS, and the package manager cant download some packages: everytime is different package missing, when I click the link on the terminal it works in the browser every single time. I have updated the channels (16.09) but I'm still experiencing these errors. Is this common?
batdog has joined #nixos
eacameron has quit [(Remote host closed the connection)]