<gchristensen>
well just keep using that foro nowo
nix-gsc-io`bot has quit [Client Quit]
<Myrl-saki>
gchristensen: Actually, you're right.
<Myrl-saki>
gchristensen: git checkout and git reflog.
<gchristensen>
sure
<Myrl-saki>
The only times where I learn how to use git. :P
Rusty1_ has joined #nixos
<srk>
what's cool is that you can even use git bisect on nixpkgs
<srk>
debugging stuff is way easier when everything is in one repo instead of making changes in number of packages/repositories like with other distros
<dtz>
yes! I used it just the other day
<Myrl-saki>
srk: ++. And git bisect is also pretty cool. To be more exact, logarithms are. 4 billion commits? You can analyze it in 30 steps!
<dtz>
$ git bisect run nix-build -A linuxPackages.perf
phaebz` has quit [Remote host closed the connection]
<Myrl-saki>
dtz: Is nix-build more "lightweight" than nixos-rebuild(I know, apples to oranges). What I mean is that if I use nixos-rebuild, I have to get every package in existence, so a git bisect wouldn't be useful.
<dtz>
oh absolutely :D
<srk>
Myrl-saki: it's still useful, hydra caches older stuff as well (not sure for how long)
<dtz>
nixos-rebuild evaluation takes significantly longer--it's trying to build a whole system! :)
<dtz>
esp when using "-A" to nix-build it's quite a bit quicker
<dtz>
but even so bisecting with nixos-rebuild probably isn't so bad
<dtz>
esp since once you automate it you can leave it, it's the ability to trivially automate it that is powerful
<Myrl-saki>
(Btw, this is on a stable branch.)
<dtz>
so taking 5min vs 15min isn't quite the point :) (err made up numbers)
<dtz>
oh, and if on stable it's all in cache so really shouldn't take very long :D
<Myrl-saki>
dtz: Right, but in my case I had to download 2 GB recently. Maybe related to Spectre+Meltdown?
<dtz>
oh, maybe.
<Myrl-saki>
Hmmm
<Myrl-saki>
Wait, let's continue on on that topic.
<Myrl-saki>
if I'm on the stable branch, is it a good idea to continually pull?
<Myrl-saki>
The reason why I no longer use unstable is because I'm no longer a fan of bleeding. But a pull in stable is semantically different from a pull in unstable.
nkpart has quit [Quit: Connection closed for inactivity]
<Myrl-saki>
The criteria for "is it a good idea" is that "I'm not a fan of huge system rebuilds nor bleeding edge"
fragamus has joined #nixos
<dtz>
stable really should always have things in cache, for what it's worth
<dtz>
and definitely should be, you know, stable :)
<dtz>
never had regrets updating stable
<dtz>
occasionally a problem on unstable but TBH those are rare or already fixed on staging ;)
<Myrl-saki>
dtz: Pfft. Master is the best.
<joepie91>
master is for when you hate idle CPU cores
<joepie91>
:P
<Myrl-saki>
joepie91: Because of all the perl scripts running?
* Myrl-saki
ducks
<joepie91>
lol
<MichaelRaskin>
master is for when you love idle CPU cores and what to give them something interesting to do!
<joepie91>
MichaelRaskin: that's... one perspective
<romildo>
I have lost the efi partition of my noteebok by accident. How can I reconstruct the systemd-boot entry for NixOS?
<fearlessKim[m]>
Ericson2314 are you here ? wanna ask a question for a possible kernel PR :>
mbrgm has quit [Ping timeout: 256 seconds]
mbrgm has joined #nixos
<hyper_ch>
romildo: one option is to reboot into live system, mount everything as it should under /mnt and run nixos-install again
stepcut has joined #nixos
xcmw has quit [Ping timeout: 268 seconds]
nocoolnametom_ has quit [Remote host closed the connection]
<romildo>
hyper_ch, "nixos-install --chroot" followed by "nixos-rebuild boot" would work?
MP2E has joined #nixos
dag__ has joined #nixos
Supersonic112 has quit [Disconnected by services]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] nicknovitski opened pull request #34126: init: kubectl at 1.7.9 (alias for kubernetes with only kubectl component) (master...kubectl) https://git.io/vN2KJ
<dhess>
Has anyone here gotten the gitlab NixOS test fixture to succeed? Mine *always* times out, it appears that the initial migrate step runs forever
<dhess>
oh it appears the test is commented out in nixos/release.nix :(
mk-fg has joined #nixos
<dhess>
so maybe I should ask instead, any tricks to speeding up disk IO on the NixOS virtualbox test fixtures?
mk-fg has quit [Changing host]
mk-fg has joined #nixos
swayf has quit [Remote host closed the connection]
<hyper_ch>
geez, it took me like 10 min to figure out what's wrong with this cronjob entry: 0 0,6,12,18 root /root/jlBU
<hyper_ch>
I shouldn't have gotten up so early
xcmw has joined #nixos
<hyper_ch>
dhess: why vbox and not kvm/qemu?
<dhess>
hyper_ch: maybe it's kvm. So speeding up virtual disk IO on NixOS test fixtures, then.
s33se_ has joined #nixos
<hyper_ch>
dhess: you don't know if it's vbox or kvm?
<dhess>
No
<dhess>
There's nothing special about my config so it's whatever NixOS's test fixtures use by default
<gchristensen>
ah that is qemu, using kvm if your host supports it
s33se has quit [Ping timeout: 248 seconds]
<dhess>
my host does support kvm. ok cool so that's even better than virtualbox. Anyway, looking for ways to speed up disk IO on this gitlab NixOS test
<hyper_ch>
virtio with kvm has good speeds
<dhess>
anyone know whether virtio is enabled in the NixOS test fixtures?
<dhess>
actually I'm looking at the gitlab.nix preStart script and I'm wondering if it's doing a bunch of unnecessary work
<fearlessKim[m]>
running `nix-build -A linux_mptcp --arg 'localSystem' 'let lib = (import <nixpkgs/lib>); in lib.recursiveUpdate (lib.systems.elaborate { system = builtins.currentSystem; }) { platform = test-platform; }' '<nixpkgs>' --show-trace` fails. I have test-platform defined in my overlay, how can I solve this ?
dkao has quit [Quit: WeeChat 2.0]
<elvishjerricco>
Is there some bash thing I can use to replace all the symlinks in $out pointing to the `tmp` directory with the actual files from there?
kp__ has quit [Quit: WeeChat 2.0.1]
<hyper_ch>
elvishjerricco: what's your actual goal?
<ekleog>
Hey! Just wondering, does anyone know of a good way to debug a big structure in nix? I tried builtins.trace (toString mydata) (...); but I'm hurt by the “cannot coerce a set into a string” thing...
<elvishjerricco>
hyper_ch: I have a build that produces a bunch of symlinks to the build directory. I want to install the directory tree that has these symlinks into $out, but obviously I can't have symlinks to the build directory
<joepie91>
gchristensen: anything more modern/recent as well by any chance?
<fearlessKim[m]>
joepie91: blender ?
<gchristensen>
I don't know, I don't do much with guis... :)
<joepie91>
ding ding ding we have a winner
<joepie91>
okay, going to mark this down as a success then :P
<samueldr>
joepie91: blender-ish gui?
<joepie91>
my objective was to approximate the design of Blender input elements in CSS
<joepie91>
that appears to have succeeded :P
<samueldr>
though I haven't really used blender
<joepie91>
thanks for all of your input btw :P
<samueldr>
now, that's static, do they behave fine in use?
<joepie91>
hehe, there's no interaction yet
krey has quit [Ping timeout: 264 seconds]
<joepie91>
samueldr: there's a project I'm working on that requires a rather information-dense UI, and for all Blender's UI faults, it does that part quite well - so I'm intending to use this as a base to design my own UI off
<joepie91>
so I'm not too concerned about interaction yet
<joepie91>
right now I'm just writing composable style mixins
Arcaelyx has joined #nixos
<joepie91>
anyway, I'm quite surprised at how far I could get with CSS, to be honest - there's absolutely no images involved in that screenshot, it's purely CSS
<samueldr>
I can believe it, there's so many ways to cheat!
<joepie91>
hehe
<joepie91>
samueldr: the trick is a gradient, a dark border, and two inset box shadows; one for the dark edge at the bottom, one for the highlight at the top
<joepie91>
the shadows are really subtle but have a big impact
<samueldr>
that's definitely what I would have suggested or guessed, shadows and gradients are useful, even for non-shadows and non-gradient uses
<jeromelanteri>
it seems taht there is 3 different UEFI boot options: systemd-boot, grub-efi, gumiboot. I'm searching a (non-obsolete) link who clearly teach things existing about that for install nixos.
hakujin has quit [Ping timeout: 246 seconds]
<jeromelanteri>
hyper_ch, thank you, i see your link with options. where are the general explications and examples around that ?
<hyper_ch>
I don't believe in efi so no idea
hakujin has joined #nixos
<jeromelanteri>
hyper_ch, it is yur choice to not believe in actual technologies, i have no intention to talk about your choices in your life.
<jeromelanteri>
If someone could provide some links for humble users who want to install nixos without need to have a master or high level study about EFI or who doesn't want to test and pain and loose time with that... thank you
asuryawanshi has joined #nixos
<samueldr>
jeromelanteri: any specific thing missing or you need to know?
<samueldr>
(I'm gathering missing bits of knowledge to then writeup a complete EFI guide)
hakujin has quit [Ping timeout: 256 seconds]
<samueldr>
AFAIK, gummiboot options are all aliased to systemd-boot options
<jeromelanteri>
samueldr, and yes, all of this (lloks like ther eis a lot when i write: "all of this"...) is stile very confused to me when i read the manual (by the way, all the other part i read are fine, but this part specifically stile too short and confused.
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<jeromelanteri>
and i can see that this point has been pointed from 2 years and half ago already. Maybe someone who know (it should have a lot) could improve the manual by share his knowledge ?
<jeromelanteri>
if people have config and don't know really or can not explain whythis works... just add exemple of what is currently working should be a start after 2 years and half...
<jeromelanteri>
why not ?
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
elninja44 has joined #nixos
<jeromelanteri>
CMCDragonkai added nice comment at the end who point on lake of information in the manual also...
<samueldr>
an EFI setup with NixOS isn't that different than for a legacy boot setup, the main bullet points are making sure the ESP is mounted at the right location and enabling one of the two efi bootloaders
<samueldr>
though, what is lacking is non-nixos-specific explanation on how this all fits in (in my opinion)
<samueldr>
looking at CMCDragonkai's comment, the last two points are not specific to EFI I think
<samueldr>
point 1 (multiple disks) I don't know, someone with experience with software RAID would know
<jeromelanteri>
samueldr, your opinion doesn't care about opinion of non-specialist user who just want to install nixos and said that for them, it is stile not clear. An opnion ha snot an obligation to denie other opinions... and the best for nixos is to be accessible, not to be more agnostic. (i'm not talking about the last two points... and i never said that his comment desribe all the problem too... i said that his comment is also nice
<jeromelanteri>
for try to make devs and manual redactors of nixos in situation to make nios more accessible wise)
<samueldr>
maybe there's someting lost in translation, but I was saying that "yes, documentation needs to be improved"
<jeromelanteri>
samueldr, do you like the idea to make nixos more accessible and manual more clear ?
<jeromelanteri>
hyper_ch, i think you also not understand what i writed... i'm not a specialist, i need your help.
<samueldr>
point 2 (size) there's no good answer, nixos can make the ESP disk usage grow quickly when keeping around older generations
<jeromelanteri>
hyper_ch, do you want to help ? or not ?
<jeromelanteri>
so on my side, (and i am not representative of all newbie corner users, but i 'm thinking that may questions should be in the column of "newbies corner"...)
<samueldr>
and ESP size also needs to take in consideration other systems that will use it, as they too can add up quickly
<jeromelanteri>
from the documentation manual, the 2 first points are clear for me
<hyper_ch>
jeromelanteri: as I said, feel free to improved the nixos wiki
<jeromelanteri>
has long as i let 256 Mo for EFI system
<cmcdragonkai>
I'd suggest doing 1 GiB ESP that's what i have
<cmcdragonkai>
and the evry once and a while do your GC and make sure to reboot
<cmcdragonkai>
so old stuff gets cleared
<jeromelanteri>
hyper_ch, you reapet yourself (it server nothing), try to first understand what i said please.
<hyper_ch>
jeromelanteri: I do understand you but you don't understand me
<cmcdragonkai>
i forgot the resource for what is needed to force a cleanup somehow
<jeromelanteri>
cmcdragonkai, hi, did you try to df on your partyition after installation for see how much place is really needed ?
<MP2E>
the most difficult part of an EFI install for me was just getting the install media to boot :P
<cmcdragonkai>
no cause that's not what really happens
<MP2E>
after that it was fairly straightforward
<cmcdragonkai>
every nixos rebuild which touches the bootloader portions
<cmcdragonkai>
involves a new generation
<cmcdragonkai>
that means a new initrd and new kernel (possibly)
<jeromelanteri>
hyper_ch, ok so it serve nothing to speak together, our discution stile close. thank you hyper_ch
<cmcdragonkai>
and if you're like me, I have ZFS in my initrd
<cmcdragonkai>
so my initrd is pretty big
<hyper_ch>
I dream of a world where I can have all that stuff as seperated datasets in a zfs pool without needing to create pre-fixed partitions
<cmcdragonkai>
so it fills up
Mic92 has quit [Ping timeout: 256 seconds]
<cmcdragonkai>
i dream of a world where UEFI supports zfs
<cmcdragonkai>
never oging to happen hahaha
ryanartecona has joined #nixos
<hyper_ch>
uefi should die IMHO
<jeromelanteri>
i tryed zfs... this file system like to use CPU a lot. i feel this fs like something for big data speciale case storage.
hakujin has joined #nixos
<samueldr>
is grub's support for zfs enough to keep all the files on the ZFS filesystem while keeping a minimal grub on ESP?
<jeromelanteri>
not for home use
<jeromelanteri>
never mind...
<hyper_ch>
samueldr: I tried it but couldn' get it to work
<jeromelanteri>
so the dificulty for me is this part:
<hyper_ch>
samueldr: with legacy, not efi
<jeromelanteri>
Instead of boot.loader.grub.device, you must set boot.loader.systemd-boot.enable to true. nixos-generate-config should do this automatically for new configurations when booted in UEFI mode.
<hyper_ch>
samueldr: but maybe I just did it wrong :)
<jeromelanteri>
so what about other options ?
<jeromelanteri>
what about other possible choices ?
<samueldr>
I wouldn't be surprised if grub's implementation diverged or wasn't updated to support all zfs features (baseless assumptions)
<jeromelanteri>
You may want to look at the options starting with boot.loader.efi and boot.loader.systemd-boot as well.
<jeromelanteri>
ok... why not be more clear on it right onow in this place of the manual ?
<hyper_ch>
samueldr: feel free to test it in a vm :) and let me know when you succeed :)
<jeromelanteri>
After having mounted your installation partition to /mnt, you must mount the boot partition to /mnt/boot.
<samueldr>
(I know absolutely nothing about ZFS! and don't have much time to spend playint with it even though it probably is a find piece of tech)
<jeromelanteri>
so that is clear... definitly not same as other distribution, not on /mnt/boot/efi then ?
<samueldr>
exact jeromelanteri, nixos assumes the ESP is mounted at (/mnt)/boot
<jeromelanteri>
yes, that point is clear
<jeromelanteri>
the two other (from your link exactly), one up and one down, are not (for me).
<jeromelanteri>
I need more precision
<jeromelanteri>
if you are able to help me, i promise i will edit this manual for improve it
<jeromelanteri>
hyper_ch, (happy)
<jeromelanteri>
so questions...
<jeromelanteri>
if i just use boot.loader.systemd-boot enable = true; i can remove all the other options ?
<jeromelanteri>
what's happen after that ? it is systemd who get boot loader charge job ?
<jeromelanteri>
If yes, is it possible to multi-boot with that ?
<jeromelanteri>
i think... it should have two big section inside "UEFI Installation": 1/ one boot 2/ multi-boot
<jeromelanteri>
then inside each one, separate for different technics able and possible to use with nixos
<jeromelanteri>
coudl you (would you) help to improve this way ?
hakujin has quit [Ping timeout: 246 seconds]
<jeromelanteri>
samueldr, ?
<samueldr>
I haven't yet looked at systemd-boot, so I can't answer for sure
<jeromelanteri>
samueldr, ok, nop. which technic for uefi boot do you choose (and how many there is ?)
<jeromelanteri>
i'm very happy also you try to help me to clarify the situation... thank you
<samueldr>
I use grub2, and AFAIK, it is the only other supported option for UEFI boot and NixOS
<jeromelanteri>
so on different distro (archlinux) i use grub2 also on EFI. It should be fine.
<samueldr>
I haven't setup dual-boot in NixOS, I'm using a separate bootloader on my only computer with another OS, which loads NixOS's grub install
<jeromelanteri>
samueldr, coud you pastebin your configuration.nix config file please ?
<jeromelanteri>
or a similare exempel on that boot point ?
<jeromelanteri>
samueldr, maybe we can deliberatly choose to first reolve one step each first... so mono-boot for grub2 uefi (forst step)
<jeromelanteri>
first
hakujin has joined #nixos
<jeromelanteri>
(actually i tryed legacy bios boot from Qemu... easy as soon has it is largely know and well documented)
<jeromelanteri>
so i'm testsing from qemu
<jeromelanteri>
so ok... stay fixed on the ball... one boot grub2 uefi method...
endformationage has quit [Quit: WeeChat 1.9.1]
<jeromelanteri>
samueldr, ? you would like to share a part of your config file and explain each option ? or not ?
<samueldr>
I'm sorry, I won't be able to help right now as I'm heading to bed, when I setup I initially was pretty familiar with ESP, grub and general linux boot process, looking at https://nixos.org/nixos/options.html#boot.loader.grub. then all made sense, that's about the best I can do now, sorry
<jeromelanteri>
samueldr, ok, nop, you tryed to help, better than do nothing or cover the problem. thank you and good night.
<jeromelanteri>
cmcdragonkai, so for this part of doc, it should be fine to add precision on that... then how much place exactly for one config with zfs did you really use ? (again, with "ncdu command tool, it is very easy to see by command line)
<jeromelanteri>
?
schoppenhauer has joined #nixos
<copumpkin>
can I `nix-env -e "foo" -f bar -iA baz`?
<copumpkin>
i.e., remove one thing and simultaneously add another
elninja44 has quit [Remote host closed the connection]
hakujin has quit [Ping timeout: 265 seconds]
hakujin has joined #nixos
babyflakes has joined #nixos
knupfer has quit [Ping timeout: 256 seconds]
hakujin has quit [Ping timeout: 256 seconds]
Arcaelyx_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hakujin has joined #nixos
evhan has quit [Ping timeout: 268 seconds]
evhan has joined #nixos
Rusty1_ has quit [Quit: Konversation terminated!]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] sellout opened pull request #34130: git-hub: add darwin support (master...fix/git-hub-darwin) https://git.io/vN2DN
NixOS_GitHub has left #nixos [#nixos]
rcshm has quit [Ping timeout: 246 seconds]
mightyby1e has left #nixos [#nixos]
a6a3uh has joined #nixos
hakujin has quit [Ping timeout: 265 seconds]
hakujin has joined #nixos
fragamus has joined #nixos
hakujin has quit [Ping timeout: 246 seconds]
hakujin has joined #nixos
<Li[m]>
can I use `builtins.readFile` on a shell file and use nix 'substitutions' in it?
<adelbertc>
there used to be a link in the Topic of this channel for NixOS swag
<adelbertc>
anyone remember what that is?
<fearlessKim[m]>
for the GSoC, Nixups is not on the list ? that would be my favorite project. Getting sthg like home-manager/nixups in nixos
rauno has joined #nixos
ma27 has joined #nixos
<fearlessKim[m]>
my bad it appears in the issues
ryanartecona has quit [Quit: ryanartecona]
Mateon2 has joined #nixos
Mateon1 has quit [Ping timeout: 248 seconds]
Mateon2 is now known as Mateon1
ma27 has quit [Ping timeout: 276 seconds]
hakujin has quit [Ping timeout: 246 seconds]
JosW has joined #nixos
hakujin has joined #nixos
MercurialAlchemi has joined #nixos
rct has joined #nixos
MP2E has quit [Remote host closed the connection]
hakujin has quit [Ping timeout: 268 seconds]
mbrgm has quit [Quit: ZNC 1.6.5+deb1 - http://znc.in]
hakujin has joined #nixos
kreetx has joined #nixos
ylwghst has joined #nixos
hakujin has quit [Ping timeout: 246 seconds]
hakujin has joined #nixos
a6a3uh has quit [Quit: Bye]
a6a3uh has joined #nixos
hakujin has quit [Ping timeout: 256 seconds]
asuryawanshi has joined #nixos
a6a3uh has quit [Client Quit]
hakujin has joined #nixos
<rauno>
Question: i created a new package, everything build and installed correctly but the new binaries ain't in system path? Any idea how to resolve it ?
asuryawanshi has quit [Ping timeout: 256 seconds]
<srhb>
rauno: What's the package, how did you install it?
<rauno>
with make install
<rauno>
in installPhase
<srhb>
Setting the destination to $out?
<rauno>
it's a ceph package
<rauno>
yep
marusich has quit [Quit: Leaving]
<srhb>
Does the store path look correct? As in, does it have a /nix/store/xxx-ceph-yyy/bin ?
<rauno>
when i run the full path of the binary in nix/store it actually works
<srhb>
Okay, and how did you install the actual package? Did you add it to environment.systemPackages?
<rauno>
nope, as i'm just creating it, i tested the installation with nix-env -f '<nixpkgs>' --install newceph
<srhb>
rauno: And does it have a /bin path?
<rauno>
need to rename it when it's all good
<rauno>
yep
<rauno>
added $out to cmake installation path
<srhb>
Okay, and you mentioned "system path" but did you actually mean that it's not in your profile path?
hakujin has quit [Ping timeout: 246 seconds]
<srhb>
As in ~/.nix-profile/bin
<rauno>
um
<rauno>
for example, old ceph returns for which rbd: /run/current-system/sw/bin/rbd
<srhb>
Since you didn't add it to environment.systemPackages, there's no reason it should be in /run/current-system/sw/bin
<rauno>
so it can be ran with just rbd
<rauno>
ah, okay
<rauno>
so it should be happening magically if i define it in nix.conf ?
<srhb>
Do you mean configuration.nix ?
<rauno>
jep
<srhb>
Yes.
<rauno>
sorry for the shorthand
<rauno>
cool, will try!
<srhb>
It's not just shorthand, there's a nix.conf that configures nix, and that's the wrong file :-)
<srhb>
(/etc/nix/nix.conf)
<rauno>
ahh okay :D
<rauno>
the package manager conf
<srhb>
Right.
* Havvy
read "conf" as conference. >_>
<srhb>
Havvy: Heh. .conf files are rare in nixland.
<srhb>
I suppose it should really be called .ini (which is even stranger)
MP2E has joined #nixos
asuryawanshi has joined #nixos
<MP2E>
anybody having issues with steam on nixpkgs unstable? I'm on latest nixpkgs master, and it's giving me this error: "/home/cray/.local/share/Steam/steam.sh: line 755: /home/cray/.local/share/Steam/ubuntu12_32/steam: No such file or directory"
<MP2E>
I checked and the file it is looking for exists and the right permissions are set
<MP2E>
not sure what's going on there
<MP2E>
(oh yeah and steam used to work a couple of weeks ago on this machine, so it shouldn't be a configuration issue)
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vN279
<NixOS_GitHub>
nixpkgs/master 246ba46 Vladimír Čunát: Merge older staging (finished on Hydra)...
NixOS_GitHub has left #nixos [#nixos]
<srhb>
silver_hook: Welp, guess your only bet is to drop libreoffice for now (and maybe get it from unstable, I think it's good there) assuming that's the one pulling in that dependency
spear2 has quit [Remote host closed the connection]
<srhb>
silver_hook: I'm unsure if a backport of something was missed, I can check laster.
<srhb>
laster*
<srhb>
...
<srhb>
later
Itkovian has joined #nixos
<rauno>
ha, good point srhb :
<rauno>
ha, good point srhb :)
ertes-w has joined #nixos
Tucky has joined #nixos
<srhb>
Hmm, maybe that's not the dependency at all..
hakujin has quit [Ping timeout: 246 seconds]
<rauno>
srhb, is there any advantages having different outputs for lib/bin/doc etc ?
<rauno>
outputs = [ "out" "lib" "bin" ]; for example
MP2E has quit [Ping timeout: 248 seconds]
<srhb>
rauno: Yeah, it reduces size needed, since for instance if you're using a lib output, but not the bin output, the bin output can be removed entirely.
<rauno>
this is there place where i could pkgs.ceph.lib for some other package as dependency ?
<rauno>
this the place*
<srhb>
rauno: Could you rephrase? I didn't understand that.
a6a3uh has quit [Quit: Bye]
<rauno>
i mean that for exampele if libvirt required one library from ceph then i don't have to put whole ceph as it's build dependency and just use it's libraries ?
<srhb>
Right :)
<rauno>
okay, this nix package world is getting more and more interesting :D
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat deleted yegortimoshenko-patch-1 at d98f4d7: https://git.io/vN25B
<rauno>
ah another question, is wrapProgram always needed ?
<rauno>
or only when there's some path issues ?
<etu>
If you can patch all the paths that the program uses (like bash-scripts and such), you probably don't need to wrap it.
<rauno>
ok cool
<rauno>
looks like nix auto replaces at least some of them
hakujin has joined #nixos
<lejonet>
srhb: well I've just got 1 headache to fix and the service module should be possible to use :P
<lejonet>
rauno: The ceph libs are conveniently accessible if you use ceph.lib in your nix expression :)
<rauno>
yep, saw that
<lejonet>
:)
<rauno>
check the link above about your repository, does your 12.2 install without problems ?
<rauno>
i checked*
<adisbladis[m]>
rauno: You get automatic patching of shebangs. wrapProgram is used when something is required in $PATH
<lejonet>
Yeah, I've gotten it compile and install just fine, I've not yet tested the binaries fully tho, as I've not gotten the conf part done completely yet
<adisbladis[m]>
(Ok some more cases too, but pretty much $PATH)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti opened pull request #34134: Update our default version of OpenSSL from 1.0.x to 1.1.x (master...openssl-1.1.x-update) https://git.io/vN2dn
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
I managed to get both 10.7 and 12.2 to compile without any issues
<rauno>
lejonet, wow, just wondering that you still had this autogen.sh in your code but current version of ceph doesn't have it anymore
<lejonet>
rauno: Well the actual nix package I've hardly do anything to, honestly :P
<rauno>
ahh, okay :
<rauno>
ahh, okay :)
<lejonet>
s/do/done/
a6a3uh has joined #nixos
<lejonet>
Probably going to clean it up a tiny bit before doing a PR, but as it works, it works :P
<srhb>
lejonet: Awesome! :D
hakujin has quit [Ping timeout: 246 seconds]
asuryawanshi has joined #nixos
<rauno>
Hum okay, i'll keep and eye on your PR then
<rauno>
just got my 12.2 package also working with libvirt and qemu integration
<lejonet>
srhb: all I've got it whining on now is my usage of if cfg.mon.enable then map (systemd.services."ceph-mon@${mon}".enable = true) cfg.mon.daemons; its not liking the if there
<rauno>
lejonet, will you mind adding the libvirt and qemu options for ceph also?
<lejonet>
rauno: once I've gotten ceph fully up and running that was going to be my next task actually, because I'm also going to integrate with libvirt and qemu :)
<rauno>
cool +1
hamishmack has joined #nixos
<rauno>
then i'll switch to your clean package in future, mine is quite messy as i needed it fast
<lejonet>
srhb: so my thought is using (abusing?) lib.optional for that, instead of if
<lejonet>
hehe :)
<silver_hook>
srhb: I’m not in a hurry with the upgrade (other than perhaps Spectre/Meltdown mitigation on my laptop)
<srhb>
lejonet: Sure, it would appear you're missing an else. :?
<srhb>
lejonet: Same deal.
<srhb>
silver_hook: OK. Sorry, I can't look into it further before I get off work.
<lejonet>
srhb: hmm, is that a requirement? I don't want an else... (I guess I could just add else "", but that is uuugly! :P )
<srhb>
lejonet: It's not ugly, it's The Right Thing.
SOO7 has quit [Quit: Quit]
<silver_hook>
srhb: No problem. Thanks for looking into it :)
<lejonet>
srhb: then The Right Thing is ugly ;)
<srhb>
lejonet: Consider foo = if bar then true
SOO7 has joined #nixos
<srhb>
lejonet: What's foo if bar is false?
<lejonet>
srhb: well for an variable assignment it makes sense to require it to have for both options, but now I don't assign it anywhere, I just want a map run conditionally
<srhb>
Same thing!
asuryawanshi has quit [Ping timeout: 256 seconds]
<lejonet>
No! (yes, I see your point ;) )
<srhb>
:-)
hakujin has joined #nixos
<lejonet>
I'll add it when I get home, hopefully I can get a cluster running this evening, so I can test setting up mds and radosgw too
burgdorfer has joined #nixos
<lejonet>
rauno: if you have a link to your messy package, that shows what you did for the qemu/libvirt integration, that would be helpful :)
<srhb>
lejonet: yay! :D Looking forward to trying it out
<lejonet>
Tho my guess is that the integration is actualy enabled on qemu/libvirt rather than ceph :)
<rauno>
my ceph package is a mess but for libvirt and qemu it was simple
<rauno>
libvirt needs confiugre flag: "--with-storage-rbd" and qemu: "--enable-rbd"
<lejonet>
srhb: Finally had some energy to work on it some more this weekend, so now I've just gotten it whining on the ifs, then it should atleast run the service file, lets hope that the conf toINI generates looks the way I think it does
SOO7 has quit [Client Quit]
<rauno>
+ ceph package as build input
SOO7 has joined #nixos
<srhb>
lejonet: Sounds great!
<lejonet>
rauno: as I thought, I'm guessing its "just" ceph.lib they actually need? Should probably make that an optional dep (yep, I'm still in gentoo mode when it comes to that :P ) so that you don't have to pull in ceph as standard for regular libvirt/qemu
hakujin has quit [Ping timeout: 255 seconds]
<rauno>
lejonet, should be yep
a6a3uh has quit [Quit: Bye]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rauno>
wow, even libvirt storage pools now connect with ceph with 12.2 version
<lejonet>
rauno: sweet, I'll put that in my notes and see to it when I'm don :)
<lejonet>
done*
<rauno>
sure, if you run into problems with ceph package you can ask :D
<rauno>
went through most of the build issues
asuryawanshi has joined #nixos
<rauno>
al thou i have just tested the rbd part of it
<lejonet>
Sounds good, I'm still a bit of a ceph novice, only done some cursory tests with their ansible build scripts a year or so back :)
dywedir has quit [Remote host closed the connection]
dywedir has joined #nixos
fragamus has joined #nixos
<lejonet>
srhb: Went and added the code now, but can't test it till I get home, but now I won't forget to add it when I get home atleast :P
<srhb>
lejonet: \o/
<lejonet>
srhb: however, the service module should be fully functional now
<srhb>
That's exciting! I can't wait to give it a spin :)
lonokhov has joined #nixos
<lejonet>
Only thing I actually need to figure out the nix part for is the setup of radosgw's
<srhb>
Let me know if you need a hand some evening :)
<lejonet>
Will do :) I'll certainly want vict... ehm, testers ;)
<srhb>
:D
<lejonet>
Especially as the only 2 parts I'm interested in with ceph is the regular mon/osd parts and the mds part for cephfs
NixOS_GitHub has joined #nixos
fragamus has quit [Ping timeout: 276 seconds]
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vN2Fi
<NixOS_GitHub>
nixpkgs/master a2b8264 Tuomas Tynkkynen: linux_testing: 4.15-rc8 -> 4.15-rc9
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
I don't really care for radosgw, but atleast want the service module to be flexible enough to allow for configuring that too
Kingsquee has joined #nixos
<lejonet>
(tho, from what I've managed to read from the ceph homepage, radosgw's are conf'd as clients anyway, so the only thing that I need to add to the service module is a way to enable startup of radosgw daemons, which should be easy enough)
hakujin has joined #nixos
civodul has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
<srhb>
I've only ever ran radosgw with the default (nonexistant?) config.
<srhb>
And added users through the web API or with radosgw-admin
hakujin has quit [Ping timeout: 256 seconds]
<burgdorfer>
Hi all, I'm trying to use nix-ops to import my default.nix and deploy to EC2, but the following error is thrown: "error: attempt to call something which is not a function but a set". What does this error mean? My default.nix uses (...).project(...), can it not be imported then by other scripts?
<lejonet>
That is 100% more than I've ever done with it :P
leat has quit [Ping timeout: 268 seconds]
zzamboni has joined #nixos
hakujin has joined #nixos
<burgdorfer>
Never mind, solved it by not calling the imported default.nix build expression. (removed {})
<rauno>
has anyone ran into provlem where fetchgit hash has changed ?
<rauno>
was able to install package on one system fine, and on another it failed cause the hashes didn't match ?!
zzamboni1 has joined #nixos
zzamboni has quit [Remote host closed the connection]
goibhniu has joined #nixos
<lejonet>
Corrupt download on the other system?
<rauno>
not sure :|
<lejonet>
I would try again, otherwise you might have a problem :P
hakujin has quit [Ping timeout: 276 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 7369530 Tuomas Tynkkynen: patchelfUnstable: Fix hash
<NixOS_GitHub>
nixpkgs/master 630a26a Tuomas Tynkkynen: git: 2.16.0 -> 2.16.1
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 2 new commits to master: https://git.io/vN2NY
NixOS_GitHub has left #nixos [#nixos]
alex`` has joined #nixos
Itkovian has quit [Remote host closed the connection]
dan_b has joined #nixos
hotfuzz_ is now known as hotfuzz
Itkovian has joined #nixos
periklis has joined #nixos
magnetophon has quit [Remote host closed the connection]
<joko>
Hello, I would like to keep some derivations in nix store even after nix-collect-garbage, how to do so in NixOS?
<joko>
More or less I want to keep stuff like stage-2-init.sh in the nix store, so that nothing gets downloaded if I modify configuration files, e.g. network settings
<LnL>
you can create your own gcroot
leat has joined #nixos
<LnL>
the -d flag only removes old profile roots not other things like indirect roots
<LnL>
eg. nix-build '<nixpkgs>' -A hello --out-link ~/hello-root
frankqux has joined #nixos
frankqux_ has quit [Ping timeout: 246 seconds]
<joko>
LnL: what if I need to build a complete nixos system? E.g. I see dependencies such as gnumake, paxctl, stage-2-init.sh, stdenv and patchelf getting downloaded and then removed on the next "nix-collect-garbage -d"
hakujin has joined #nixos
kreetx_ has joined #nixos
kreetx has quit [Ping timeout: 246 seconds]
<manveru>
rust on nixos-unstable seems broken :(
<LnL>
link?
<manveru>
trying to build again, but it takes like an hour
<LnL>
joko: nix-instantiate '<nixos/nixos>' -A system --indirect --add-root /etc/nixos/system-root
<joko>
sphalerite: gc-keep-outputs did it, but it keeps outputs for every derivation, right? Is there an option to keep outputs just for the nixos-system one?
ckauhaus has joined #nixos
<sphalerite>
joko: I don't think so but no it doesn't keep outputs for every derivation
<sphalerite>
Only for the derivations that have GC roots attached to them
knupfer has quit [Ping timeout: 264 seconds]
<sphalerite>
So it does keep outputs for every derivation remaining in the store, but not every derivation remains in the stire
ckauhaus has quit [Remote host closed the connection]
<ixxie>
If a commit includes work by multiple people, is there a convention for appending something to commit like: co-authored with: Jane Doe <jane.doe@example.org>
<NixOS_GitHub>
nixpkgs/release-17.09 c9d1a42 Ioannis Koutras: clion: 2017.3.1 -> 2017.3.2...
<therp>
did somebody toy around with retpoline already? (gcc 7.0 git branch)
ssmike has joined #nixos
ssmike has quit [Remote host closed the connection]
ssmike has joined #nixos
hakujin has joined #nixos
ckauhaus has quit [Remote host closed the connection]
hakujin has quit [Ping timeout: 255 seconds]
aloiscochard has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 2 new commits to master: https://git.io/vNaUt
<NixOS_GitHub>
nixpkgs/master 95880aa Tuomas Tynkkynen: nixos/initrd: Don't include some x86-specific modules unconditionally
<NixOS_GitHub>
nixpkgs/master 4432699 Tuomas Tynkkynen: build-support/vm: Autodetect kernel filename...
NixOS_GitHub has left #nixos [#nixos]
<aloiscochard>
hi all, does any one know how I can avoid having this warning everytime I start a shell?
<aloiscochard>
"warning: Nix search path entry ‘/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs’ does not exist, ignoring "
<aloiscochard>
I don't understand why it is throw.
periklis has quit [Ping timeout: 268 seconds]
KingOfThePirates has joined #nixos
<KingOfThePirates>
Hello! I'm trying to install NixOS on a partition via rEFInd. My only OS right now is Windows. I'm a bit lost in the documentation. I was wondering if someone might have a few spare minutes to make sure I'm on the right track
jtojnar_ has joined #nixos
jtojnar has quit [Ping timeout: 248 seconds]
jtojnar_ is now known as jtojnar
krey has joined #nixos
<infinisil>
aloiscochard: what's the output of `nix-info`?
<KingOfThePirates>
I'm not at the part where I have nix yet :P
<KingOfThePirates>
I successfully installed NixOS on a VM guest
<infinisil>
(i asked aloiscochard)
<KingOfThePirates>
Ah haha
hakujin has joined #nixos
<infinisil>
KingOfThePirates: I have installed NixOS on a previously only Windows machine though
<infinisil>
KingOfThePirates: do you have some free space for nixos?
<ixxie>
infinisil: well, I read that, but since none of the solutions is perfect I was wondering if Nixpkgs has a convention about which to use
TweyII has joined #nixos
TweyII is now known as Guest54122
mkoenig_ has joined #nixos
hakujin has quit [Ping timeout: 256 seconds]
mkoenig has quit [Ping timeout: 240 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to master: https://git.io/vNakV
<NixOS_GitHub>
nixpkgs/master 7dd50de Vladimír Čunát: knot-resolver: 1.5.1 -> 1.5.2 (security)...
NixOS_GitHub has left #nixos [#nixos]
<ixxie>
I keep getting an error of "Module X has unsupported attribute `Y`; this is caused by assignments to the top-level attributes `config` or `options`."
a6a3uh has quit [Quit: Bye]
<ixxie>
Y is services and system, in different attempts
<infinisil>
ixxie: what's your nix expression?
<ixxie>
infinisil: hmm... nevermind somehow I resolved it >.<
johann__ has joined #nixos
Phillemann has joined #nixos
<Phillemann>
I'm trying to pin nixpkgs in my nix derivation. I download nixpkgs via "src = fetchFromGithub { ... }" and then access packages via "pkgs = import src {};"
<Phillemann>
However, in my derivation, there's <nixpkgs/nixos/...>. How do I make that relative to my downloaded nixpkgs?
<Phillemann>
Just "src/nixos/..."?
<infinisil>
Phillemann: import "${src}/nixos/..."
<Phillemann>
I guess I'm not sure what this thing I download via fetchFromGithub "is".
<infinisil>
Phillemann: it's a nixpkgs checkout, the same as <nixpkgs> (which expands to something like /path/to/nixpkgs)
<Phillemann>
Okay, but in the nix language. Is it a set? A string?
<Phillemann>
"${src}..." seems to work btw., thanks infinisil.
<infinisil>
fetchFromGitHub { ... } is a derivation, and putting it in a string like "${fetchFromGitHub { ... }}" makes it expand to the derivations outPath, something like /nix/store/...
<Phillemann>
Ah, I see.
Ivanych has quit [Ping timeout: 240 seconds]
<KingOfThePirates>
infinisil: Thank you for your help. I'm going to see what I can do
Ivanych has joined #nixos
KingOfThePirates has quit [Ping timeout: 260 seconds]
<MoreTea>
python3 + pypi2nix + cffi + nix question: if I want my C code to link to libxyz, should it be sufficient to provide it as a buildInput like (import ./requirements.nix{}).mkDerivation { buildInputs = [ libxyz ]; }
<MoreTea>
I'm getting a 'ld: cannot find -llibxyz' error right now.
<MoreTea>
Is this is local error, or might it als indicate that I need to provide more libs?
<MoreTea>
ok, stupid me. I had to remove 'lib' from the name in the python code.
asuryawanshi has quit [Remote host closed the connection]
fyuuri has joined #nixos
asuryawanshi has joined #nixos
Thra11 has quit [Ping timeout: 246 seconds]
hakujin has quit [Ping timeout: 255 seconds]
ssmike has quit [Remote host closed the connection]
<aloiscochard>
infinisil: yes it started to happen since the last time I switched from stable to unstable (~1month ago), I have three computer, two of them where I did that have the issue.
<aloiscochard>
the last one, is running with same configuration but on stable, this one don't have the issue.
xcmw has joined #nixos
pkill9 has joined #nixos
<infinisil>
aloiscochard: no idea what could cause this. This file is missing right?
<infinisil>
s/file/folder
leat has quit [Ping timeout: 246 seconds]
tomberek has quit [Ping timeout: 260 seconds]
<aloiscochard>
infinisil: there is a `nixos` and `pkgs` folders there, but no `nixpkgs`. So yes, it is missing for real.
dan_b has quit [Ping timeout: 268 seconds]
<aloiscochard>
might be related to some of my configuration or user profile... I'll continue the hunt, thanks for your help anyway infinisil
<infinisil>
hmm, well nixpkgs is usually just a symlink to .
johann__ has joined #nixos
<aloiscochard>
oh that's good to know.
<infinisil>
aloiscochard: hold on, why do you have 2 times the same channel?
<infinisil>
just noticed that in your nix-info output
hakujin has joined #nixos
<aloiscochard>
right! must be a typo!
<aloiscochard>
I'll try cleaning that first, let see.
<aloiscochard>
this is very weird, so there was duplicate channels, one as `nixos` the other as `nixpkgs`. I did remove the later, the run `channel update`.
elurin has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
<aloiscochard>
Now I can not rebuild the system or open a nix-shell, I'm getting "error: file ‘nixpkgs/nixos’ was not found in the Nix search path (add it using $NIX_PATH or -I" /cc infinisil
strobelight has joined #nixos
<infinisil>
huh
<aloiscochard>
I remember entering that channel with wrong name (nixpkgs instead of nixos) at the time of moving from stable to unstable.
pkill9 has left #nixos ["WeeChat 2.0.1"]
<infinisil>
aloiscochard: what's your NIX_PATH?
hakujin has quit [Ping timeout: 276 seconds]
dan_b has joined #nixos
<aloiscochard>
infinisil: here it is `nixpkgs=/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs:nixos-config=/etc/nixos/configuration.nix:/nix/var/nix/profiles/per-user/root/channels`
<aloiscochard>
should I be worried about that `nixpkgs` prefix here?
<infinisil>
aloiscochard: i mean you could just stop using nix-channel (which I'm doing), and use a git checkout instead
<infinisil>
but that's just avoiding the problem
<aloiscochard>
I tried the opposite (kkeping the channel with nixpkgs removing the one with nixos), and I'm getting "‘/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs"
<aloiscochard>
infinisil: right I use to do that, when in need of hotfix
<aloiscochard>
thanks for trying anyway
<infinisil>
aloiscochard: getting that for what?
burgdorfer has joined #nixos
<aloiscochard>
infinisil: sorry wrong copy past, I have `error: file ‘nixpkgs/nixos’ was not found in the Nix search path `
leapingfrogs has quit [Quit: WeeChat 1.9.1]
<burgdorfer>
My nix-deploy keeps getting stuck right after downloading grub, any ideas what could be going on or how I could look into this further? --debug does not give me any more information.
<infinisil>
burgdorfer: can you show the output of a run?
ssmike has quit [Remote host closed the connection]
a6a3uh has joined #nixos
<infinisil>
aloiscochard: yup
<aloiscochard>
infinisil: you saved me :) okay that was just one small warning... but it was making my shell ugly for weeks now. Thanks a lot for your time! that was it...
<jekor>
Is there a hook for extracting the source when running nix-shell on a .drv (in the nix store)?
<fearlessKim[m]>
jekor: post/preUnpack
<burgdorfer>
infinisil: Absolutely, one second. Also, running nixops deploy again afterwards results in "copying closure ..." instead, which also does not seem to progress. strace tells me it keeps calling select and not progressing.
<jekor>
Thanks. They don't seem to exist for this though. It's a Haskell package if that makes a difference.
leat has joined #nixos
<jekor>
(I don't see them in the environment.)
<ixxie>
anybody got an example of an nginx virtual host with the sslCertificate option set? somehow having trouble getting a pilfered module working
<ij>
Can I somehow find out whether some (in this case hardware.cpu.intel.updateMicrocode) option's been enabled in nix-repl?
<burgdorfer>
After downloading grub 100% it just sits there forever, no CPU or network activity (at least 30 minutes, after which I stopped it)
nix-gsc-io`bot has quit [Client Quit]
johann__ has joined #nixos
<gchristensen>
burgdorfer: maybe just try again?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master b2d0d04 Tuomas Tynkkynen: ddccontrol: Mark x86-only...
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vNaZN
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
aloiscochard: glad i could help :)
ylwghst has quit [Remote host closed the connection]
ssmike has joined #nixos
<infinisil>
burgdorfer: yeah, if you haven't already, just try again, seems like a network connectivity hickup
a6a3uh has quit [Quit: Bye]
<infinisil>
ij: Options are enabled for a nixos system, not a nix-repl
<pauldub>
Hi everybody, I can't build rustc 1.17 on nixos, one of the tests fails: `[run-pass] run-pass/out-of-stack.rs` is it a known issue?
<gchristensen>
pauldub: it is a known issue, being worked on
<pauldub>
Ok, thanks!
<infinisil>
ij: you can check options in your current configuration (the one defined in configuration.nix), with `(import <nixpkgs/nixos> {}).config.hardware.cpu.intel.updateMicrocode`
<ij>
infinisil: That is, of course, clear.
hakujin has joined #nixos
<ij>
infinisil: So where should I write that?
<gchristensen>
that one you can write in the nix repl
<infinisil>
ij: nix-repl works, or nix-instantiate --eval -E '(import ...'
<ij>
thanks!
<sphalerite>
ij: or just nixos-option hardware.cpu.intel.updateMicrocode on the command line
hke has quit [Quit: bye]
<infinisil>
sphalerite: does nixos-option incorporate your configuration.nix?
<ij>
Seems so, yes.
<ij>
It looks like nixos-generate-config didn't add that, though it should've.
hke has joined #nixos
<sphalerite>
infinisil: yes, it even tells you where the value of the option is set
<burgdorfer>
gchristensen, infinisil: I've tried several times all throughout the day and have not experienced any other networking issues. However, I do believe the same configuration did successfully deploy this morning, but then failed due to a build error. I don't think I've modified the deployment config since then, hence my confusion.
<sphalerite>
infinisil: of course it has a bunch of other problems like not being able to handle options with .s in their names
hakujin has quit [Ping timeout: 246 seconds]
<gchristensen>
infinisil: ...what?
<infinisil>
gchristensen: huh? did i miss something?
<gchristensen>
sphalerite: ^ I mean, d'oh. *makes coffee*
<infinisil>
burgdorfer: nix shouldn't download something it has already downloaded, so if you run nixops deploy again, all these fetches shouldn't be present
<infinisil>
ah
<sphalerite>
gchristensen: if one of the elements of the attrpath of the option has a dot in it, like the services.nginx.virtualHosts.* often has, there's no way to query it using nixos-option
<sphalerite>
AFAIK
<gchristensen>
oh! I see, yes that makes some sense to me
<burgdorfer>
infinisil: If I stop it when it is stuck and then run it again it says "copying closure ..." and then just hangs there forever.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 3d7dad7 adisbladis: Merge pull request #34129 from samueldr/dbeaver/4.3.3...
<NixOS_GitHub>
[nixpkgs] adisbladis pushed 2 new commits to master: https://git.io/vNack
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master d9031c6 Samuel Dionne-Riel: dbeaver: 4.3.2 -> 4.3.3
chisui has joined #nixos
<burgdorfer>
Is it even supposed to copy closure at that point or is something terribly wrong?
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
a6a3uh has joined #nixos
<infinisil>
burgdorfer: no that's correct, it first fetches everything it can get from the cache, then copies the rest from your build-machine to the deploy-machine
<gchristensen>
it is supposed to, it means it is copying files from your system to the target system which were not possible to fetch from the binary cache
<gchristensen>
infinisil: jinx
<infinisil>
heh
<gchristensen>
infinisil: I'll let you keep going, you're doing great and I don't need to interrupt :)
<infinisil>
well i have no idea what's the problem
iyzsong has joined #nixos
ssmike has quit [Remote host closed the connection]
<infinisil>
the copying closure part usually takes a while for me too (without output)
<infinisil>
but no network activity is werid
ssmike has joined #nixos
ma271 has joined #nixos
ma27 has quit [Ping timeout: 265 seconds]
xcmw has joined #nixos
<infinisil>
burgdorfer: you could try copying the closure manually with nix-copy-closure (which is pretty much what nixops does)
kuznero has joined #nixos
<kuznero>
Hi All!
jensens has joined #nixos
<kuznero>
Does anybody have any experience setting up kubernetes cluster in NixOS while in the same time being using proxy?
<infinisil>
kuznero: just ask the question you actually have
<burgdorfer>
infinisil: Oh cool, that makes sense. I'll look into that.
<kuznero>
While enabling simple nix config with roles = [ "master" "node" ], I am getting the following error: Handler for POST /v1.32/images/create returned error: Get https://gcr.io/v2/: proxyconnect tcp: dial tcp 127.0.0.1:3128: getsockopt: connection refused
zzamboni has quit [Quit: Leaving.]
hakujin has joined #nixos
zzamboni has joined #nixos
evhan has quit [Ping timeout: 268 seconds]
<kuznero>
Sandbox perhaps?
evhan has joined #nixos
<kuznero>
Nope, turning off sandbox does not help :(
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Nadrieril opened pull request #34142: ckb: 0.2.8 -> 0.2.9 + project moved (master...update-ckb) https://git.io/vNaCz
NixOS_GitHub has left #nixos [#nixos]
hakujin has quit [Ping timeout: 246 seconds]
frankqux has quit [Quit: leaving]
a6a3uh has quit [Quit: Bye]
xcmw has quit [Ping timeout: 256 seconds]
<ixxie>
kuznero: probably a stupid remark, but is the port open?
<kuznero>
Yes, I am running cntlm proxy that listens on 3128. And it works when I am just using the system in a normal way. But when I am trying to `nixos-rebuild switch` it seems that it is running in a different context so that it cannot reach 127.0.0.1:3128.
<gchristensen>
I bet MoreTea could help, kuznero
<kuznero>
gchristensen: what is MoreTea?
a6a3uh has joined #nixos
<infinisil>
lol
pie__ has joined #nixos
<gchristensen>
MoreTea is a friendly helpful member of the nixos community who does things with the likes of kubernetes
ssmike has quit [Remote host closed the connection]
Phillemann has joined #nixos
<Phillemann>
I saw that I can get a TeX environment using texlive.combine and such, but are there tools to support building a PDF via a nix expression, so I can type "nix-build" and it'll output the PDF?
<infinisil>
Phillemann: have you looked at pandoc?
<ixxie>
Phillemann: its easy enough to do something similar with any other tex engine / wrapper
<Phillemann>
Ah, seems simple enough.
stepcut has joined #nixos
mrkgnao has joined #nixos
<ixxie>
Phillemann: as a side note, I have recently started to move towards markdown instead of using tex; less flexible but can do 90% of the common tasks
<ixxie>
Phillemann: with a huge reduction in complexity
fyuuri has quit [Ping timeout: 256 seconds]
ma271 has quit [Ping timeout: 246 seconds]
hakujin has joined #nixos
<infinisil>
ixxie: pandoc let's you use markdown and tex combined
<niksnut>
Phillemann: nixpkgs contains some functions for building latex documents in pkgs/tools/typesetting/tex/nix
<infinisil>
ixxie: although it's kinda hairy sometimes
<ixxie>
infinisil: exactly; I use tex templates and markdown content, and if I ever needed advanced features I suppose I could throw in some latex in the document itself
<Phillemann>
ixxie: I'm not a huge fan of markdown. Then again...not a huge fan of TeX eitehr.
<ixxie>
Phillemann: I like markdown, I just miss some neat and flexible layout tools there
<ixxie>
Phillemann: what is your prefered markup lang?
<Phillemann>
We're using asciidoc at work. And for personal stuff and note-taking I'm using org-mode.
<Phillemann>
Which at least has excellent ASCII tables support. :)
<infinisil>
hmm org-mode, still need to get more into it (and emacs for that matter)
<infinisil>
it can do a lot
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat pushed 1 new commit to release-17.09: https://git.io/vNa4V
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/release-17.09 fabdb3b Vladimír Čunát: knot-resolver: upstream patches for CVE-2018-1000002
zzamboni has quit [Quit: Leaving.]
alexteves has joined #nixos
<ixxie>
does org-mode have flexible tables?
i-am-the-slime has joined #nixos
<ixxie>
e.g. nestable tables
<Phillemann>
No, I don't think so.
<i-am-the-slime>
I think my nix-shell thingy where I do some python magic does not look at my ~/.pgpass flie
<i-am-the-slime>
Also how can I add something to my PATH in that nix-shell?
<infinisil>
i-am-the-slime: if you want a package, use the -p flag
<infinisil>
i-am-the-slime: but you can just add to path like normal otherwise (export PATH="new/path:$PATH")
<layus>
Is there any way to get the actual .drv file path in a nix expression
<i-am-the-slime>
What package? And where do I use the -p flag?
zzamboni has joined #nixos
<infinisil>
i-am-the-slime: `nix-shell -p hello` for having the hello package available
Rusty1 has joined #nixos
<infinisil>
layus: use the `drvPath` attribute of a derivation
<layus>
Thks !
<i-am-the-slime>
ah, I have a default.nix
<i-am-the-slime>
with a lot of packages
mizu_no__ has joined #nixos
xcmw has joined #nixos
hakujin has quit [Ping timeout: 256 seconds]
zzamboni has quit [Ping timeout: 268 seconds]
<infinisil>
i-am-the-slime: and you want to add some packages to the nix-shell without adding them to the file?
mizu_no__ has quit [Client Quit]
kuznero has quit [Quit: Page closed]
<i-am-the-slime>
No, I just want the pgpass file to be used
<i-am-the-slime>
inside of the shell
dan_b has quit [Ping timeout: 246 seconds]
nschoe has joined #nixos
peacememories has joined #nixos
<infinisil>
i-am-the-slime: well no idea, the program that uses this file will probably look at some env var, but i can't help you with that
hakujin has joined #nixos
<infinisil>
i-am-the-slime: you can set any env var for a nix-shell by just doing `export FOO=bar` in shellHook, or for all of the derivation by just setting `mkDerivation { Foo = "bar"; ... }`
xcmw has quit [Ping timeout: 256 seconds]
<i-am-the-slime>
Looks like it was a different problem. I was using the software wrong :)
mizu_no__ has joined #nixos
<i-am-the-slime>
I put it in the global bashrc, that's okay for me, thanks infinisil
<i-am-the-slime>
I also seem to have trouble with ssh-agent
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master f9025bb Tuomas Tynkkynen: crrcsim: Mark x86-only...
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vNaBb
mizu_no__ has quit [Client Quit]
<i-am-the-slime>
maybe you also know about that
<Phillemann>
Simple question about derivations: We have $out for the results of the derivation, and $src for the source files. Is the "buildCommand" then executed in a temporary directory (where I can put intermediate results from building)?
hakujin has quit [Ping timeout: 255 seconds]
<infinisil>
Phillemann: yeah, that's in a tmp dir created new for every build, you can put stuff there
<Phillemann>
Excellent, thanks.
<infinisil>
(or use mktemp if you wish)
<Phillemann>
But the contents of $src aren't copied there automatically, right?
<gchristensen>
they are
<i-am-the-slime>
The question is simply: How do I autostart ssh-agent?
<i-am-the-slime>
So what do I need to do to configuration.nix
<gchristensen>
well, they are as part of stdenv.mkDerivation and many other build helpers, Phillemann, but not as part of the lowest deepest parts of nix
<ixxie>
infinisil: so I can just refer to $tmp by default?
<gchristensen>
ixxie: I use programs.gnupg.agent.enable and programs.gnupg.agent.enableSSHSupport
<Phillemann>
gchristensen: Hm, then maybe I'm doing something stupid by using mkDerivation, but defining a buildCommand in there?
<gchristensen>
i-am-the-slime: I use programs.gnupg.agent.enable and programs.gnupg.agent.enableSSHSupport
johann__ has quit [Quit: Leaving.]
<infinisil>
gchristensen tries to complete nicks after 1 char confirmed :P
<gchristensen>
Phillemann: don't use buildCommand, it eliminates all the built-in smarts
<i-am-the-slime>
gchristensen: and that works?
<i-am-the-slime>
I found that
<gchristensen>
infinisil: :P, i-am-the-slime: it does work for me, yes
<i-am-the-slime>
and saw this line: programs.ssh.startAgent = true;
<infinisil>
ixxie: i'm not sure about $tmp, never used it
<i-am-the-slime>
Is this one mutually exclusive with the one you mentioned?
<gchristensen>
i-am-the-slime: I'm not sure :/
<i-am-the-slime>
do you also have that line?
<infinisil>
gpg ssh support is different than ssh-agent i think
<gchristensen>
i-am-the-slime: my programs.ssh.startAgent is set to false
<gchristensen>
infinisil: "Enable SSH agent support in GnuPG agent. Also sets SSH_AUTH_SOCK environment variable correctly. This will disable socket-activation and thus always start a GnuPG agent per user session. "
<i-am-the-slime>
okay, I mean I can try that one
<i-am-the-slime>
see you after a reboot
<ixxie>
infinisil: so whatever isn't put in $out ends up in the temporary directory
i-am-the-slime has quit [Quit: Konversation terminated!]
<infinisil>
ixxie: well that's pretty much all you have in your sandbox, so by necessity
i-am-the-slime has joined #nixos
<ixxie>
infinisil: right :P
hakujin has joined #nixos
i-am-the-slime has quit [Ping timeout: 256 seconds]
jrolfs has joined #nixos
i-am-the-slime has joined #nixos
<i-am-the-slime>
Nice. I think I updated KDE and now it crashes almost all the time. But this happened with an nix-env --upgrade.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vNauv
<NixOS_GitHub>
nixpkgs/master c4eb230 volth: nixos/libvirtd: add qemu-img to $PATH of the daemon...
<i-am-the-slime>
At least the ssh-agent seems to be running.
ssmike has joined #nixos
acarrico has quit [Ping timeout: 256 seconds]
<goibhniu>
i-am-the-slime: KDE often freaks out if you have any Qt packages installed in your user profile
hakujin has quit [Ping timeout: 246 seconds]
<i-am-the-slime>
okay, how do I avoid that?
<goibhniu>
you can put them in environment.systemPackages instead
<i-am-the-slime>
move them to configuration.nix?
<goibhniu>
exactly
<i-am-the-slime>
but how do I find out which ones there are?
szicari has joined #nixos
<i-am-the-slime>
also on every reboot the system starts about 15 instances of emacs, and I don't know why
<goibhniu>
something like `nix-store -q --referrers /nix/store/*-qt*` should help
<i-am-the-slime>
Is there maybe an easy way to move all packages that are currently installed through nix-env to systemPackages?
<i-am-the-slime>
Woah
ylwghst has joined #nixos
<goibhniu>
well, you can see what you have in your user profile with `nix-env -q`
<i-am-the-slime>
ah cool
<i-am-the-slime>
I guess stuff like "kmail" "kontact" "korganizer" "konversation" look suspect
hakujin has joined #nixos
<goibhniu>
yes indeed :D
<i-am-the-slime>
so I can just move the same strings to the configuration file
<i-am-the-slime>
and I guess somehow remove them with nix-env -e?
<i-am-the-slime>
and reboot
<goibhniu>
I'm not sure if a reboot is necessary, but yeah
<i-am-the-slime>
okay, let's try that
<goibhniu>
the "names" which you see in `nix-env -q` don't necessarily match the "attribute path" required for systemPackages ... so you might need to fiddle a bit
<infinisil>
Adluc: works no problem for me, it's probably your network
<Adluc>
okay, will check it
<Adluc>
thanks
sary has quit [Ping timeout: 240 seconds]
<hyper_ch>
how can one use wayland with kde in nixos?
<Phillemann>
Hm. I define "checkInputs = [ pkgs.jdk ];", which seems to download openjdk, but "javac foo.java" still tells me "javac: command not found".
xcmw has joined #nixos
<Phillemann>
(I've got a makefile with: "check: javac foo.java" in it)
alexteves has joined #nixos
<manveru>
Phillemann: you mean buildInputs?
<Phillemann>
manveru: No, I need the jdk only for testing purposes, so I thought the "checkInputs" property would be the correct one.
<manveru>
it's either buildInputs, or nativeBuildInputs, depending on your case
<manveru>
i don't think there's a checkInputs :)
<Phillemann>
checkInputs Dependencies needed for running the checkPhase. These are added to buildInputs when doCheck = true.
<manveru>
oh damn, there is
<Phillemann>
Oh, but it's below "buildPythonPackage parameters"
<manveru>
yeah
<manveru>
it's not on mkDerivation anyway
<Phillemann>
Ok, so for non-python packages, I just put those dependencies inside buildInputs.
<manveru>
exactly
<Phillemann>
Good to know, thanks.
nuncanada has quit [Quit: Leaving]
zzamboni has joined #nixos
suvash_away has quit [*.net *.split]
mujx[m] has quit [*.net *.split]
voobscout[m] has quit [*.net *.split]
scott1 has quit [*.net *.split]
sudoreboot[m] has quit [*.net *.split]
moredhel1 has quit [*.net *.split]
sargon[m] has quit [*.net *.split]
copumpkin has quit [*.net *.split]
NickHu1 has quit [*.net *.split]
regnat[m] has quit [*.net *.split]
pvn[m] has quit [*.net *.split]
smbarte2[m] has quit [*.net *.split]
moredread[m] has quit [*.net *.split]
AdamSlack[m] has quit [*.net *.split]
nickdshsadhkas[m has quit [*.net *.split]
wizek[m] has quit [*.net *.split]
adva[m] has quit [*.net *.split]
nixo[m] has quit [*.net *.split]
bobvanderlinden_ has quit [*.net *.split]
wilornel[m] has quit [*.net *.split]
wip_ has quit [*.net *.split]
thoughtpolice has quit [*.net *.split]
manveru has quit [*.net *.split]
brodul has quit [*.net *.split]
spawnthink[m] has quit [*.net *.split]
mhsjlw[m] has quit [*.net *.split]
coffeeshot has quit [*.net *.split]
EB[m] has quit [*.net *.split]
rnhmjoj[m] has quit [*.net *.split]
happyente[m] has quit [*.net *.split]
Singularity[m] has quit [*.net *.split]
yochai[m] has quit [*.net *.split]
Vincenzo[m] has quit [*.net *.split]
kitten[m] has quit [*.net *.split]
pstn has quit [*.net *.split]
techtangents[m] has quit [*.net *.split]
baconicsynergy[m has quit [*.net *.split]
Barnabas[m] has quit [*.net *.split]
M-liberdiko has quit [*.net *.split]
ycy[m] has quit [*.net *.split]
Pneumaticat has quit [*.net *.split]
magnap has quit [*.net *.split]
indefini has quit [*.net *.split]
M-berdario has quit [*.net *.split]
timclassic has quit [*.net *.split]
WilliamHamilton has quit [*.net *.split]
spacekitteh[m] has quit [*.net *.split]
grahamc has quit [*.net *.split]
Ralith has quit [*.net *.split]
Baughn has quit [*.net *.split]
andi- has quit [*.net *.split]
dmj` has quit [*.net *.split]
cocreature has quit [*.net *.split]
wavewave has quit [*.net *.split]
edofic has quit [*.net *.split]
alphor has quit [*.net *.split]
swflint has quit [*.net *.split]
ChongLi has quit [*.net *.split]
lstanley has quit [*.net *.split]
brodul has joined #nixos
hakujin has joined #nixos
ChongLi has joined #nixos
swflint_away has joined #nixos
Baughn has joined #nixos
wip_ has joined #nixos
alphor has joined #nixos
swflint_away is now known as swflint
cocreature has joined #nixos
edofic has joined #nixos
wavewave has joined #nixos
thoughtpolice has joined #nixos
Sonarpulse has joined #nixos
dmj` has joined #nixos
manveru has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
i-am-the-slime has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
bobvanderlinden_ has joined #nixos
wilornel[m] has joined #nixos
Singularity[m] has joined #nixos
happyente[m] has joined #nixos
yochai[m] has joined #nixos
rnhmjoj[m] has joined #nixos
smbarte2[m] has joined #nixos
adva[m] has joined #nixos
kitten[m] has joined #nixos
regnat[m] has joined #nixos
NickHu1 has joined #nixos
moredhel1 has joined #nixos
nixo[m] has joined #nixos
andi- has joined #nixos
mujx[m] has joined #nixos
ycy[m] has joined #nixos
Pneumaticat has joined #nixos
Ralith has joined #nixos
wizek[m] has joined #nixos
pstn has joined #nixos
sudoreboot[m] has joined #nixos
lstanley has joined #nixos
moredread[m] has joined #nixos
copumpkin has joined #nixos
nickdshsadhkas[m has joined #nixos
EB[m] has joined #nixos
voobscout[m] has joined #nixos
Vincenzo[m] has joined #nixos
mhsjlw[m] has joined #nixos
M-liberdiko has joined #nixos
stepcut has joined #nixos
hakujin has quit [Ping timeout: 256 seconds]
AdamSlack[m] has joined #nixos
sargon[m] has joined #nixos
scott1 has joined #nixos
techtangents[m] has joined #nixos
spawnthink[m] has joined #nixos
baconicsynergy[m has joined #nixos
Barnabas[m] has joined #nixos
timclassic has joined #nixos
pvn[m] has joined #nixos
WilliamHamilton has joined #nixos
indefini has joined #nixos
M-berdario has joined #nixos
magnap has joined #nixos
grahamc has joined #nixos
coffeeshot has joined #nixos
jmiven has quit [Quit: co'o]
spacekitteh[m] has joined #nixos
fyuuri has quit [Ping timeout: 256 seconds]
fyuuri has joined #nixos
ssmike has joined #nixos
jmiven has joined #nixos
mingc has quit [Quit: WeeChat 2.0.1]
detran` has quit [Read error: Connection reset by peer]
asuryawa_ has quit [Remote host closed the connection]
<infinisil>
mog: you can't, nixos doesn't back up your nix files used for building
asuryawa_ has joined #nixos
coot has joined #nixos
<infinisil>
If you want to check out your current configuration.nix, just look at the file though
<infinisil>
Or use (import <nixpkgs/nixos> {}).config.fileSystem.blabla to inspect certain options
<stphrolland>
Hi. I had already asked the question but can't find it in my notes. I have a haskell project which depends on GL. When compiling GHC indicates me I probably needs the dev packages. I find myself unable to locate either GL, or GL-dev using `nix-env -qA | grep GL`. How would you search for it ?
<hyper_ch>
infinisil: how do people know what they've installed if they used nix-env -iA ?
<infinisil>
hyper_ch: nix-env -q?
<mog>
thanks infinisil i think thats what i wanted the import line
<hyper_ch>
dunno :) I ust configuration.nix
alex`` has quit [Quit: WeeChat 1.9]
<hyper_ch>
s/ust/use/
<mog>
im trying to play with the pythonix nd read in my config
<mog>
but not sure how yet
<infinisil>
stphrolland: maybe you need mesa_glu or mesa
<lejonet>
Is there really no way of executing code that isn't an variable assignment in a service definition?
<infinisil>
lejonet: what are you referring to? I have no idea what you speak of
<lejonet>
infinisil: look at my link earlier, row 331-333, nixops deploy is whining on me either of unexpected if or if I try to do it unconditionally, it whines that it didn't expect a ( there :P
<lejonet>
I'm basically trying to make the service itself enable the relevant daemon services, using the instantiated version of the service, instead of having the user do it themselves (and thus maybe forgetting and then wondering why no daemon was started)
digitus has joined #nixos
ma271 has joined #nixos
xcmw has joined #nixos
<lejonet>
It seems to bark at me for not assigning those maps to something, if I try to do it inside the config = {}, which I can understand, as that is just a set
mkoenig_ has quit [Remote host closed the connection]
<lejonet>
but moving them outside of it doesn't work either, so I'm thinking that its not supposed to be possible to do something like that with a service definition, which kindof sucks even tho I understand
mkoenig has joined #nixos
fyuuri has quit [Ping timeout: 260 seconds]
ssmike has quit [Ping timeout: 276 seconds]
<infinisil>
lejonet: um, there are only 330 lines?
<lejonet>
haha yeah, I just pushed a commit that removed them to test if the ini generation works as I hope, let me find the correct url based on the commit instead of latest
shabius has joined #nixos
<infinisil>
ah, i can just look at the commits then
<infinisil>
it just merges multiple attrsets in configuration.nix style
shabius_ has quit [Client Quit]
<lejonet>
Hmm, won't that just change the place it whines at, as those lines technically don't fit into an attrset at all, as they don't set any value at all?
<lejonet>
(that is the maps return value is not stored anywhere, because its not relevant)
<infinisil>
don't they set the systemd.services."ceph..." services?
<infinisil>
nix doesn't "execute" anything btw, it just declares stuff and uses that stuff elsewhere
<lejonet>
Well yeah, but what would that return in the map?
shabius has quit [Ping timeout: 246 seconds]
<lejonet>
I know it doesn't I'm using the term lazily because cba to define exactly what executes it (and honestly I don't know that much of nix and such yet to even define that)
<infinisil>
how about `mkMerge (map (mon: { systemd.services."ceph-mon@${mon}".enable = true; }) cfg.mon.daemons)`
<lejonet>
worth a try :)
<lejonet>
Because all I want to accomplish is that once the user has enabled a daemon and set which daemons should be started (the list in cfg.mon.daemons), the service should loop through that list and create an instanced service for each daemon
<infinisil>
yeah that's what that does
ckauhaus has joined #nixos
<lejonet>
if this works, I'm going to be very happy, been trying to google and scoure the docs for doing this, to no avail :(
civodul has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
<lejonet>
just a sanity check, doing a map on an empty value (i.e. cfg.mon.daemons being a empty list) won't throw a hissy fit? If it does, I'm guessing I must guard it with an mkIf then?
zzamboni has quit [Quit: Leaving.]
<Guest54122>
lejonet: It doesn't, of course
<Guest54122>
You can experiment in nix-repl
<Guest54122>
I mean, if you pass an empty list as the function then it's not going to be very happy :þ
ckauhaus has quit [Ping timeout: 256 seconds]
<lejonet>
well, giving it an empty list makes it return just an empty list it seems (map that is)
<infinisil>
lejonet: mkMerge on an empty list will output an empty set, everything is fine
<lejonet>
infinisil: yay :D
<lejonet>
I'm so happy that git rebase exist, so I can hide all these pointless commits :P
<srhb>
lejonet: I keep a personal copy of your branch that auto updates every five minutes.
<srhb>
lejonet: (Not really, it's cool. :P)
ssmike has joined #nixos
<lejonet>
srhb: xD
xcmw has joined #nixos
<JazzyEagle>
After installing both ghc and stack as indicated, I'm getting an error that indicates "The following execuatbles are missing and must be installed: make"
ma271 has quit [Quit: WeeChat 2.0]
chisui has quit [Ping timeout: 256 seconds]
<JazzyEagle>
This isn't in the install instructions anywhere, and a search is yielding a ton of results. What package do I need to install to satisfy that requirement?
<lejonet>
Well, I'm guessing I'm being a fool and doing this wrong, because nixops deploy is still mad at me, because it was expecting a . or =, not [, or (
ma27 has joined #nixos
<sphalerite>
Anyone know if there's a way to compute hashes in the nix language? e.g. if I want to make a fixed-output derivation whose output is indeed fixed but depends on the inputs
* lejonet
facedesks and goes and gets something to drink
<srhb>
JazzyEagle: It's unclear what command resulted in that error.
<srhb>
JazzyEagle: Regardless, you probably need maketools or some part of stdenv.
<srhb>
lejonet: What's the code in question?
burgdorfer has joined #nixos
<avn>
JazzyEagle: packages: [gnumake] then, but I curious, what haskell lib require it
ylwghst has joined #nixos
ssmike has quit [Remote host closed the connection]
nadley has quit [Quit: No Ping reply in 210 seconds.]
<JazzyEagle>
avn: I think ghc itself does. I know ghc was originally compiled w/ gcc, and I think it may still rely on it a bit in the background.
<burgdorfer>
Is there a way to deploy my app to ec2 and download packages on the remote machine rather than uploading them from my local host? My upload speed is quite slow and it takes forever.
<lejonet>
I'm probably doing something simple wrong, so I'm going to go and hunt for the docs for mkMerge... :P
<avn>
JazzyEagle: well, ghc itself built by make, but stack/cabal don't use make for build stuff
<JazzyEagle>
Yeah, I don't know, then, what's causing it.
<avn>
so `stack --nix build` fails about make, does it print what package it attempt to build?
<srhb>
lejonet: What's the result of mkMerge? You seem to have them just dangling as if they were instructions in an imperative language
<lejonet>
srhb: I'm guessing maybe I should move them outside of the config = {}; and do like mkMerge [ config map (bla bla) thingy ];
<infinisil>
sphalerite: just run it once and use the hash of the error message?
nadley has joined #nixos
hakujin has joined #nixos
<lejonet>
srhb: well yeah, that is kindof what I'm trying to figure out if its even possible to do with an service or if I'll have to leave that to the user to do
<sphalerite>
infinisil: no because then it'll need updating every time the inputs change
<stphrolland>
infinisil: yep that was part of the missing dependencies. It's much better with mesa, mesa_glu, freeglut, X.11 and other xorg related libs. However, now it complains that it cannot find <X11/X.h>. So I guess I don't have the headers installed, however I cannot find a pckage named libX11-dev in nixos packages. Should I search for this name, or dev packages are special in nixos, and reachable another way
<infinisil>
sphalerite: well that's the point of a fixed-output derivation
sirkha has quit [Remote host closed the connection]
<stphrolland>
?
<srhb>
lejonet: Right, this is a language misunderstanding
<srhb>
lejonet: If you meant to add them to the set, you probably want something like //
<lejonet>
The cleanest solution, from the users perspective, imo, is that when you've set ceph.mon.enable and ceph.mon.daemons is the list of daemons you want to enable, it creates those instanced services and nothing more
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34150: julia-git: mark broken, hasn't built since 2017-04-08. (master...fix/julia-git-broken) https://git.io/vNaxw
NixOS_GitHub has left #nixos [#nixos]
<raynold>
ahh it's a wonderful day
ylwghst has quit [Ping timeout: 240 seconds]
<sphalerite>
infinisil: the point of a fixed-output derivation is to allow side effects which don't affect the output
<infinisil>
stphrolland: there is xorg.libX11
zzamboni has joined #nixos
<lejonet>
srhb: yeah, it very clearly is xD
<srhb>
lejonet: Where's mkMerge from again? lib?
<infinisil>
stphrolland: what i usually do is use rg (ripgrep) on a nixpkgs checkout
<symphorien>
stphrolland: there is a tool called nix-index which can help you answer this type of questions
<sphalerite>
infinisil: I want to achieve a side effect that doesn't affect the output, but won't get reused for the same inputs. Specifically for testing derivations
<lejonet>
srhb: seems so, I've managed to find its definition in lib/modules.nix in nixpkgs repo
<sphalerite>
infinisil: I suppose I'm XYing here
rauno has quit [Remote host closed the connection]
<srhb>
lejonet: Here's what you're doing: { foo = 42; lib.mkMerge [ { bar = 42; } ] }
<lejonet>
it seems to just inherit the contenst :P
<srhb>
lejonet: Here's what you mean to be doing (probably)
<infinisil>
sphalerite: i mean, if there were something that automatically calculates the hash of a fixed output derivation, then you'd have no way to verify that the output is indeed the hash it should have. There would be no hashes in nixpkgs to verify e.g. that fetchFromGitHub did actuall fetch the right thing
<srhb>
Or merge outside that
orivej has quit [Ping timeout: 256 seconds]
<lejonet>
Well won't that just make a new set that is { foo = 42; bar = 42; }? What I'm trying to do doesn't assign or set anything meaningful to be returned from map :/
<sphalerite>
infinisil: yes there would be because the hash is still fixed, it's just a function of the build inputs
<srhb>
lejonet: I'm not sure what you are trying to do, so it's a little hard to tell. I'm just trying to clarify why it's nonsensical in nix lang, possibly failing along the way. :-)
<lejonet>
srhb: nah, I understand what you mean
shabius has joined #nixos
<infinisil>
sphalerite: but a fixed output derivation uses the hash from the output, not input
<sphalerite>
infinisil: so you can still do nix-instantiate --eval '<nixpkgs>' -A foo.outputHash
<lejonet>
and I guess I could try and do a //, that is what seems to make the least amount of non-sense in my mind xD my head is just not in functional language territory atm :P
<srhb>
lejonet: Gotcha. Are you just trying to join the attributes together?
<srhb>
I'm not sure you need mkMerge at all.
<sphalerite>
infinisil: consider a trivial example. `runCommand 'test-foo' {} ''echo ${foo} OK > $out''`
<lejonet>
srhb: You tell me... xD I'm trying to make it set the systemd service to true based on the list in cfg.mon.daemons
<srhb>
lejonet: OK, let me reread.
<infinisil>
lejonet: srhb: mkMerge should be used for configuration.nix style attrsets, it merges them correctly with stuff like lists and strings. there is also recursiveUpdate which doesn't merge stuff correctly, and // which is super primitive and doesn't even recurse into attrs
<sphalerite>
infinisil: clearly that will always produce the same output for the same foo. So you can compute its output hash purely based on foo.
<lejonet>
Basically if cfg.mon.enable = true and cfg.mon.daemons = [ "0" "1" ]; I want it to set systemd.services."ceph-mon@0".enable = true and systemd.services."ceph-mon@1".enable = true
Neo-- has quit [Ping timeout: 256 seconds]
<srhb>
lejonet: Why not something simpler like...
<sphalerite>
also neither of the wake-on-lan tools in nixpkgs seem to have the ability to specify which interface to send the wake-on-lan packet on??
<infinisil>
lejonet: did what i said earlier not work?
<infinisil>
lejonet: the config = mkMerge (map ...
michaelpj_ has joined #nixos
<lejonet>
infinisil: I'm almost 100% sure that I'm using it wrong, so no, it didn't work :P
<sphalerite>
but yes I'm XYing. Even if this existed, which I don't think it does, it wouldn't be a great tool for the job.
zzamboni has quit [Quit: Leaving.]
<srhb>
lejonet: systemd.services = (attributeset of "ceph-mon@..." = ...)
<infinisil>
lejonet: you'll need to do something like `options = ...; config = mkMerge ([ { <some common options, used all the time } ] ++ map (mon: { systemd.services."... = true; }) cfg.mons)
<lejonet>
srhb: well aren't we fancy? ;) (I had no idea how to accomplish that, so me and my imperative brain tried to do the best it could... xD )
<infinisil>
sphalerite: hmm, I think it might not be possible for it to exist, because nix wouldn't be able to guarantee it's fixed-output then
<lejonet>
infinisil: yeah, I was stupid and tried to add it inside config = {};
<infinisil>
sphalerite: since the builder script could in theory do anything
<srhb>
lejonet: Sorry, I'm really not trying to be snarky or anything. :)
aarvar has quit [Ping timeout: 256 seconds]
<srhb>
lejonet: But it seems simpler to think of assignments like that, and then you don't get stuck with "statementy things"
<srhb>
"seems" = probably more experience :P
<lejonet>
srhb: I know, I guess my sarkasm isn't spelling out properly, I'm not meaning you're snarky or all, its just my noobieness of the language and rusty programming that is collectively facepalming now xD
<sphalerite>
infinisil: but that's the case with all fixed-output derivations. It tries building it, the builder could do anything, but if the output is as expected in the end nix is fine with it
<srhb>
lejonet: OK, just making sure! :-)
ryanartecona has quit [Quit: ryanartecona]
<lejonet>
srhb: sarkasm and text is hard to mix without <sarkasm></sarkasm> tags :P
<srhb>
True true. :P I thought I read it correctly though, was just a tiny bit unsure ;-)
<srhb>
lejonet: Anyway, if you can construct that attributeset of name = { enable = true; } you've accomplished the thing you were trying with those three lines
<lejonet>
srhb: You're being very helpful, and I'm grateful for that, I just feel like an idiot for doing the first thing an imperative mind does when encountering functional languages, try to use them imperatively :P
<lejonet>
srhb: xD well the thing that makes it even more embarrassing is that even tho it was a while since I programmed, the latest programming language I fell in love with and programmed with is Elixir... which IS a functional language
<srhb>
lejonet: I know it vaguely! :)
<infinisil>
sphalerite: coming to the conclusion it should be possible
<lejonet>
srhb: so you'd think that my mind would firstly jump to functional programming and not imperative :P
chisui has joined #nixos
<srhb>
I still sometimes "bail out" into imperative when I can't envision the functional approach. That's easy in Haskell, not so much in Nix. :P
<lejonet>
srhb: its quite a nice language, but niched just like erlang, as its essentially just a modernized wrapper for erlang
<srhb>
lejonet: Oh, and if you do take the approach I suggested, you may have to restructure a bit since you can't assign systemd.services twice in the same module.
<srhb>
lejonet: But that should not be a problem.
<lejonet>
hmm, how do I get nix-repl to print out the full values, instead of truncating them like [ { ... } { ... } ] ?
rauno2 has joined #nixos
<infinisil>
lejonet: :p
<infinisil>
(not the smiley, the command :p)
<lejonet>
xD
hakujin has quit [Ping timeout: 256 seconds]
* lejonet
throws something soft, light and fussy in infinisil direction
rauno3 has joined #nixos
<srhb>
Short for :print. The smiley, followed by "rint"
rauno2 has quit [Read error: Connection reset by peer]
<lejonet>
srhb: would it be possible to create an attrset of just the "ceph-mon@${mon}" parts and merge it in at the systemd.services level? As I can do several assignments to systemd.services.somethingSomething = darkside; without it kicking up a fuss
<lejonet>
so if I have a set of { "ceph-mon@0".enable = true; "ceph-mon@1".enable = true; } couldn't I just systemd.service // daemonSet ?
<lejonet>
Kindof like I do at line 184 with the cephConfig set that gets handed to toINI?
<srhb>
lejonet: systemd.services has no value at this point, so no
<lejonet>
wouldn't even have to guard them with optionalAttrs, as map will return an empty set if the daemon list is []
<lejonet>
srhb: damnit
<lejonet>
But then I think infinisil suggestion with mkMerge would work
<srhb>
lejonet: May I just suggest a simpler approach (and infinisil can correct me if I say something wrong)
<lejonet>
soo, how do I get nix-repl to stop truncating values, and thus creating boob-like output, so that I can fiddle with map and see if I can understand how this works? (yes, that sentence can't be improved in any way to make it less... wrong :P )
<srhb>
lejonet: :print { foo.bar = 42; }
<lejonet>
I kindof want to understand how the solution works too, so I can re-use it later :P
<lejonet>
srhb: oh, good that I'm already sitting in the corner of shame :P
<srhb>
:p
* lejonet
quickly bolts for the exit
xcmw has joined #nixos
<infinisil>
lejonet: I could quickly write your ceph thing how i think it would be best if you want
<srhb>
infinisil: I'm interested at least :)
<lejonet>
infinisil: if you think you have the time, feel free :)
<lejonet>
I already owe you people a few beers/beverage of choice if I ever meet you :P
<lejonet>
Because map returns a list, so I'll have to turn that list into an attrset (iirc listToAttrs is a builtin) before merging, or just use mkMerge which seems to flatten lists and bunch em all together into one attrset
<lejonet>
So the mkMerge [ <config stuff> ] ++ map function ++ map function; is probably the easiest way, as infinisil said earlier
<srhb>
lejonet: Right, my idea involved joining the systemd.services definition using something like // listToAttrs (map (mon: { name = mon; value = { enable = true; }; } mons)
<lejonet>
But in this case I'd probably do systemd.service = mkMerge [] instead of the whole config attr
* srhb
nods
<srhb>
That would be cleaner :)
<lejonet>
srhb: I think that is what infinisil was hinting at earlier
<srhb>
Maybe, I'm waiting with bated breath ;-)
<lejonet>
Haha, you and me both
<bind>
good evening :) I have a question regarding nixos configuration ... within the by default generated configuration file there is a commented out line programs.gnupg.agent = { enable=true; enableSSHSupport=true;} is this a nice thing to have?
<srhb>
bind: That's difficult to answer. Will you be using it?
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<bind>
srhb: thats difficult to answer ... it is storing keys for gpg encryption right? what does it do in conjunction with SSH ?
<srhb>
bind: you would have to refer to the gnupg-agent documentation to get the specifics. :)
<bind>
srhb: I have no idea if I will be using it actively, would like to know more about it .. but well..
<lejonet>
bind: basically what that does is that it enables you to use a authentication subkey from a gpg key for SSH auth
<bind>
lejonet: next to the keys from ~/.ssh?
<lejonet>
by enabling gpg-agent to talk like ssh-agent (I'm using that atm, with an hardware gpgkey, nitrokey)
<srhb>
"The OpenSSH Agent protocol is always enabled, but gpg-agent will only set the SSH_AUTH_SOCK variable if this flag is given. In this mode of operation, the agent does not only implement the gpg-agent protocol, but also the agent protocol used by OpenSSH (through a separate socket). Consequently, it should be possible to use the gpg-agent as a drop-in replacement for the well known ssh-agent."
<lejonet>
bind: in theory, yes
<srhb>
That's what they say about that option.
<lejonet>
bind: you can add file keys to the gpg-agent-acting-like-ssh-agent too
<lejonet>
(but unlike ssh-agent, they only need to be added once, gpg is smart enough to remember :D )
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34157: vim: fix for cross, add missing configure test override (master...fix/vim-cross) https://git.io/vNVJq
NixOS_GitHub has left #nixos [#nixos]
<srhb>
Sheesh, *I* should be using that...
<lejonet>
srhb: its dead simple to conf
<srhb>
I'm putting off redoing my entire key infrastructure.
<lejonet>
However, it does complicate agent forwarding quite a bit
<srhb>
Because ugggh
<lejonet>
Haha :P
<azazel>
so, given a function that produces a cfg to be put at services.nginx.virtualHosts.<name> from a template, what can i do to obtain a "tree" merge of two of these sets? "//" does a flat merge, preserving just the last .virtualHosts set
<bind>
lejonet: since you are using it ... do you have some pointers for further reading or just the gpg-agent docs as srhb mentioned?
<lejonet>
bind: let me find you the yubikey docs, they are a excellent starting point
<bind>
lejonet: Thank you... and it does complicate agent forwarding quite a bit? :D
<lejonet>
bind: yes, because you can't "just" forward the agent the way you do with ssh-agent, because you really, really, don't want to expose/spread your gpg-agent that way, as it holds the private keys and such
<srhb>
azazel: We were just discussing mkMerge. :)
deltasquared has joined #nixos
<lejonet>
Basically what you have to do with gpg-agent, to forward your agent, is to create a socket forwarding (with RemoteForward) so that the traffic is sent to the actual box that has the gpg-agent running
<lejonet>
instead of copying the contents of gpg-agent and starting one at the host you want it forwarded to, with identical content
<lejonet>
(as you do with ssh-agent and agent forwarding)
<bind>
lejonet: I see ... the details are found in the docs somewhere I guess?
<lejonet>
bind: yeah, just look up RemoteForward
<bind>
lejonet: ... is that option straightforward to use? I mean .. can I just turn it on an "grow" into it?
<srhb>
azazel: There's also recursiveupdate for simple things
<infinisil>
lejonet: srhb: Done, but untested (as I don't feel like writing the config for this :P), just needs to finish pushing
<srhb>
infinisil: Cool!
<lejonet>
i.e. the way I do it is having RemoteForward /home/lejonet/.ssh-agent /run/user/1000/gnupg/S.gpg-agent.ssh, which maps my local unix socket at /run/user/1000/gnupg/S.gpg-agent.ssh to the remote unix socket at /home/lejonet/.ssh-agent
<deltasquared>
hmm, for the systemd.packages option, is it sufficient for the target package to have e.g. "$storepath/lib/systemd/system/*.service" in order for them to be copied into a profile?
<infinisil>
It was a bit more difficult than i thought
<lejonet>
bind: I think its fairly straightforward to use yes
<lejonet>
bind: if you know how to use the -L and -D options (the socks forwarding for SSH), its very easy to understand
<infinisil>
deltasquared: i think anything in $out gets linked
<bind>
lejonet: well I used those flags a while ago..
<bind>
lejonet: is the agent used for other things too? such as the kde wallet?
<deltasquared>
infinisil: what, including $out/bin? seems a bit overbearing for merely requesting it be present for systemd
<lejonet>
bind: I don't know how the KDE wallet works, but if it can use gpg to auth, it should use the gpg-agent yes
peacememories has joined #nixos
jb55 has quit [Read error: Connection reset by peer]
<lejonet>
The 3 subkeys I have on my nitrokey is afterall signing, encryption and authentication, and its only the last one that is actually used for ssh, the first 2 are used in pure gpg situations
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master f851457 Will Dietz: vim: fix for cross, add missing configure test override...
<srhb>
azazel: Yes, they're generally documented in comments near their definitions
jb55 has joined #nixos
simendsjo has joined #nixos
<lejonet>
well, always keep one way or another to either reset or remember your passphrase/pin for GPG smartcard and you can recover from any screw ups :D (I recently had to recover from failing to input the right user pin on mine... xD )
<bind>
lejonet: I can use it without a smartcard right? :D
<bind>
lejonet: just to make sure
<bind>
lejonet: :P
<lejonet>
bind: yeah, on-file key is standard mode :)
<NixOS_GitHub>
nixpkgs/master 360b86e Will Dietz: vim: 8.0.1257 -> 8.0.1428
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 2 new commits to master: https://git.io/vNVTv
NixOS_GitHub has left #nixos [#nixos]
<srhb>
lejonet: That's essentially what I had in mind (minus the mkMerge which I'm honestly not sure does much but gather up the mkIfs)
<infinisil>
lejonet: srhb: My internet is back!
<bind>
by the way ... after reading a bit on the issue tracker it seems that credential storage is a bit of an issue right now (with nix ... nixstore). It won't be a problem with the machine I am currently installing but I would like to know if it is really that troublesome as it sounds ... :D
<infinisil>
Alright now to push my code
<deltasquared>
infinisil: my understanding of systemd.packages had been that it would merge the systemd units from them into the resulting profile, is this incorrect?
<srhb>
infinisil: Woo!
ckauhaus has joined #nixos
<lejonet>
infinisil: haha, wuhu!
ckauhaus has quit [Read error: Connection reset by peer]
ckauhaus has joined #nixos
<srhb>
lejonet: The unaligned maps hurt! Why would you do that. :D
<lejonet>
srhb: in this context, just doing // and then doing a listToAttrs I guess is equivalent
ckauhaus has quit [Remote host closed the connection]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 closed pull request #34150: julia-git: mark broken, hasn't built since 2017-04-08. (master...fix/julia-git-broken) https://git.io/vNaxw
NixOS_GitHub has left #nixos [#nixos]
<bind>
srhb: so the nix store is read only for the world except for root of course (nix daemon) ... and it stores secrets?
<srhb>
bind: It's easy to make a solution like that, mostly, but each time you do it you bring nice nix attributes that easily render at build time into some shoddy runtime interpretation.
<srhb>
bind: It should not store secrets, but it does. And it's world readable, hence the issue.
<srhb>
lejonet: Yes, a fairy survived this time. Close shave!
<srhb>
:)
ckauhaus_ has joined #nixos
astsmtl has joined #nixos
astsmtl has quit [Changing host]
astsmtl has joined #nixos
<lejonet>
srhb: as long as I fix it before the PR, so it doesn't get commited and print in stone? ;)
<srhb>
Right!
szicari has quit [Ping timeout: 240 seconds]
szicari_ is now known as szicari
<srhb>
infinisil: gist? :)
<lejonet>
I'm very happy for git rebase all of a sudden xD
<infinisil>
um, hopefully gist works
<bind>
srhb: as far as I could see the issue is on the table for a while now and some encryption "solutions" were mentioned here and there ... are those the things you mean by nix attributes that render into some shoddy runtime interpretation?
<infinisil>
the reason irc has no trouble is because it's all through 1 tcp connection i think, and my internet connection only has trouble establishing /new/ tcp connections
<srhb>
bind: Or maybe you have a single user system and you don't care. Or every user is trusted and you don't care.
<srhb>
bind: I mean I want to be able to generate, at runtime, an encrypted store path (or *something* that is protected enough that only root can read it) and then have it unlocked at runtime (when the systemd unit needs it, and only for that unit)
ckauhaus_ has quit [Remote host closed the connection]
<srhb>
Er, I want to be able to generate at BUILD time.
<lejonet>
infinisil: and the fact that irc is a really, really, lightweight protocol that hardly cares about latency nor bandwidth :P
<bind>
srhb: my laptop is basically single user programming laptop :D ... but if it fares well I would like to use some of the things learned elsewhere of course
<lejonet>
srhb: kindof like nixops keys and friends?
<srhb>
lejonet: Well, yes in that the secret is only available to the service and at runtime.
<srhb>
But kept in Nix for as long as possible otherwise, wwithout begin world readable.
oida has quit [Quit: WeeChat 1.6]
<lejonet>
infinisil: the (mkIf ) around the service files kindof renders the = mkIf part unneccessary :P
mrkgnao has quit [Ping timeout: 248 seconds]
<infinisil>
lejonet: how so?
<infinisil>
ahh
<infinisil>
yeah
<infinisil>
can remove the second mkIf, didn't notice that
<lejonet>
infinisil: :)
<srhb>
Uh, does systemd parameterization even work like that
<srhb>
There's no longer a ceph-mon@ service, only ceph-mon@foo1, ceph-mon@foo2
<infinisil>
the second mkIf would be kinda wrong anyways, because then it would still set `systemd.services."ceph-mds@${name}" = {}`
<lejonet>
The ceph-mon@ service has to exist for ceph-mon@0 to exist
<lejonet>
because ceph-mon@0 is just an symlink to ceph-mon@
* srhb
nods
<infinisil>
oh
<lejonet>
That is why I've made the ceph-<daemon>@ unconditionally but the instances of them conditionally
<infinisil>
hold on, i totally ignored the systemd @ stuff
<infinisil>
never used that
<lejonet>
^^
<srhb>
Ah :)
<bind>
great ... that manual encrypted root install didn't pan out too well :D ... "waiting for device to appear ...." rinse and repeat
<lejonet>
Fair enough, I've only come across it a few times
<lejonet>
Its not that common
<stphrolland>
infinisil: just for your information, the X11 headers file are located in the xorg.xproto packages. I'm still blocked in the build, but on another missing file. So I guess xorg.xproto is the good one.
ckauhaus has quit [Ping timeout: 256 seconds]
<infinisil>
Alright but my approach still works, just put the ceph-<daemon>@ stuff in a mkIf cfg.<daemon>.enable block :p
<srhb>
bind: Well, it's absolutely workable. The PR you mentioned is a result of a situation where it was unacceptable that gitlab's database password (for instance) was world readable on the host.
* bind
will try to be quiet now ... until he shudders of dispair ..
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/vNVkr
<NixOS_GitHub>
nixpkgs/master 64fa24b Ioannis Koutras: xmonad-log: init at 0.1.0
<lejonet>
infinisil: ^^ you're a lisper? (judging from all those parantesis... ;) )
<srhb>
bind: Hopefully some day a better solution will materialize that doesn't butcher the systemd services completely just to do runtime interpolation :)
<infinisil>
lejonet: heh not really, but yeah it reminded me of that :P
<lejonet>
infinisil: nah, if cfg.<daemon>.daemons is an empty list, map returns an empty list :)
<srhb>
Oh, and that..
<lejonet>
and the ceph-<daemon>@ are conditionally created with the mkIf
<infinisil>
lejonet: i mean the ceph-<daemon>@foo stuff in the maps
<infinisil>
they get created even if <daemon>.enable = false
<infinisil>
which might not have an effect
<infinisil>
because the @ main thing isn't enabled
<infinisil>
but it still puts them in the configuration which isn't very nice
<srhb>
That's true. At least if the systemd parameters work like that.
<srhb>
Or parameterized units.
<lejonet>
well unless both <daemon>.enable = true AND <daemon>.daemons != [], nothing should happen
<lejonet>
atleast that is what I hope... :P
phdoerfler has joined #nixos
<infinisil>
well, if somebody was to create a systemd.service."ceph-<daemon>@" himself, has <daemon>.enable = false, and <daemon>.daemons != [], then something unexpected would happe
<lejonet>
map (mon: { systemd.services."ceph-mon@${mon}".enable = true; } ) daemons in nix-repl with daemons = []; returns []
<infinisil>
lejonet: but if it's not [], it unconditionally enables these services, regardless of whether <daemon>.enable = true
<bind>
hmm .. I am following the guide over here https://gist.github.com/martijnvermaat/76f2e24d0239470dd71050358b4d5134 ... and I figured that I need to add the boot.initrd.luks.devices section to my configuration ... but there is a device uuid in there so that should go into hardware configuration?
<srhb>
infinisil: Isn't that unavoidable anyway? You can't prevent a user from creating foo@bar if it doesn't already exist.
<lejonet>
infinisil: indeed, that is why I've been pondering adding a optional for each map (that is what I had initially, but if cfg.<daemon>.enable then map stuff;)
<infinisil>
lejonet: yeah that's what I'd add
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] arcadio opened pull request #34159: libreoffice: Add Spanish (es) language support and sort langs list (master...libreoffice-es) https://git.io/vNVLO
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
infinisil: Good to have someone else also check my work :D
<infinisil>
srhb: yeah, not the problem I was trying to get at though
* srhb
nods
<lejonet>
(I am serious about the owing both infinisil and srhb a few beers/beverage of choice if I ever meet them irl :) )
* bind
will just try it out..
<srhb>
lejonet: I'm closer, I think, so beware. ;)
<infinisil>
lejonet: NixCon 2018 maybe :P
<lejonet>
srhb: oh noes!
<lejonet>
infinisil: where is that?
<infinisil>
no idea
<srhb>
I don't think it has been announced, if there even is one.
<infinisil>
not known yet
<lejonet>
would be an ideal place for it tho
<lejonet>
depending on how far it is, I might go
<srhb>
lejonet: Where are you located again?
<lejonet>
srhb: The icy depths of sweden
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
And I learned about the @ systemd stuff today :D
<lejonet>
infinisil: \o/
<deltasquared>
infinisil: think you can help me out with my query then? :P
<lejonet>
infinisil: run!
<infinisil>
deltasquared: which query?
<deltasquared>
just trying to understand what systemd.packages actually does - I can find it's definition but not where it's used.
* infinisil
has a look
<deltasquared>
grah, why am I so tired today, it's not even 8PM yet
<infinisil>
Ahhhh
<deltasquared>
?
<typetetris>
seems there is no xorg log file with nixos ?
<lejonet>
srhb: is it now I should be afraid of being stalked? :O
<lejonet>
Its not me misaligning stuff, this time tho!
<srhb>
I'm off, have a good evening! And nice work so far :)
<lejonet>
srhb: thanks for all the help :)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c closed pull request #34159: libreoffice: add Spanish (es) language support and sort langs list (master...libreoffice-es) https://git.io/vNVLO
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
O.o "value is a set while a list was expected" I guess I shouldn't use mkMerge that way afterall
<lejonet>
guess that makes sense tho, having a list of sets
<lejonet>
for constructing all services, easier to loop over and all
mahalel_ has left #nixos [#nixos]
ryanartecona has joined #nixos
<bind>
someone in here with encrypted lvm root? :D
<lejonet>
not with nixos :P
<bind>
pff
<bind>
:P
<bind>
would like to compare and match .nix configuration..
<lejonet>
I run encrypted btrfs rootfs on several systems, but that is with own written initramfs and gentoo :P
orivej has joined #nixos
<bind>
hmm
<deltasquared>
infinisil: thanks for tracking the implementation of that option down anyway.
<deltasquared>
now, I must part, lest I fall asleep and my head hits the desk
deltasquared has left #nixos ["*headdesk*"]
zzamboni has joined #nixos
<bind>
now you mention it .. I got side-tracked yesterday about encryption and disk encryption .. do you use a usb device (or your yubikey e.g.) for unlocking the system? and do you have a unencrypted boot partition on the system itself?
<bind>
lejonet: ^
<bind>
wildly interesting topic ..
<lejonet>
bind: varies, on the desktops I have it, I've gone the lazy route and just have an unecrypted boot partition with some weak security for the keys but for my laptop I have an USB device as the boot device that contains everything
<phdoerfler>
my nixos wants to compile rustc and fails doing so and I don't know why it wants rust in the first place. How do I stop this?
<bind>
lejonet: but the root fs is on the laptop I guess?
<lejonet>
bind: yeah, would be odd otherwise :P
<infinisil>
Damnit, i was away
<bind>
lejonet: puppy linux like and the laptop is just some storage medium with peripherals .. :D
<infinisil>
couldn't answer delta in time that i just did a `rg systemd.packages` on nixpkgs
JosW has quit [Quit: Konversation terminated!]
<lejonet>
infinisil: :P
<lejonet>
bind: indeed :D
<bind>
lejonet: whish I could get that to work .. grrr .. the boot process keeps waiting for the disks to become available but they dont and I have no idea how to proceed :D
<bind>
lejonet: maybe I should also just go with gentoo hahaha
<lejonet>
bind: that is where the custom initramfs comes into play :D
* lejonet
has a dislike for autogenerated initramfses due to that
<infinisil>
phdoerfler: it's a known problem and according to gchristensen is being worked on
<infinisil>
can't find an issue about it though
<maurer>
phdoerfler: No idea about a fix, but most likely the reason it wants rust is for firefox
<bind>
lejonet: I guess it would take some time to become proficient with that?
ThatDocsLady has quit [Ping timeout: 276 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/staging a786218 Daiderd Jordan: Merge branch 'master' into staging
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 1 new commit to staging: https://git.io/vNVmh
<phdoerfler>
infinisil: ok thanks
<lejonet>
bind: nah, its actually quite simple honestly, as long as you already know your head around creating static binaries (or know how to copy in all the deps of a dynamic one)
JazzyEagle has left #nixos ["WeeChat 2.0.1"]
<phdoerfler>
infinisil: I hate this "wtf is my nixos doing" moments
<phdoerfler>
*these
<bind>
lejonet: it seems to work! :D finally
<lejonet>
bind: yay! :D
<bind>
lejonet: now I only have to figure out how to get a usable "enter passphrase" prompt later on ... really not high priority :D
ylwghst has joined #nixos
<bind>
lejonet: that's also a thing a custom initrd is handy for ? :D
stepcut has quit [Remote host closed the connection]
<lejonet>
Well, depends on what you mean with a usable passphrase dialogue, you can control when its showed yes, but you're usually stuck with the one that cryptsetup gives you, unless you feel like writing a wrapper for it that feds the input to cryptsetup or so :P
hamishmack has quit [Quit: hamishmack]
mahalel_ has joined #nixos
<bind>
lejonet: well ... when I boot up I guet sometimes the prompt (enter passphrase) along with some output from udev
<lejonet>
That could potentially be helped with a custom initramfs, but I just use keyfiles because I'm lazy :P
<bind>
lejonet: I would only like to have the prompt readable but I will have a look at that later on .. sounds interesting by the way ... some bootable usb stick with riced up unlock screen :D
<lejonet>
xD
<bind>
lejonet: so cryptsetup is configured to look for keyfiles on usb there?
<lejonet>
yeah
<bind>
for the lazy systems I mean
<bind>
:P
<lejonet>
A initramfs is "just" a tarball of a skeleton system, the only required parts is that you have some /dev entries (null, console, tty, random, urandom and zero) and a /init that is a binary of some sort, or a script (but then you need to have an interprenter too, I just use busybox sh)
mahalel_ has quit [Ping timeout: 256 seconds]
spear2 has joined #nixos
<lejonet>
My init is a simple shellscript that sets up /proc, /sys and /dev, unlocks rootfs with cryptsetup and mounts rootfs somewhere, cleans up and does exec switch_root /mnt/root /path/to/real/init
stepcut has joined #nixos
<lejonet>
ofc there is a rescue_shell function that installs busybox sh, with tools, incase something fails, so I can troubleshoot, but an initramfs isn't anything more magical than that :P
ckauhaus has joined #nixos
<catern>
hmm, what other reasons are there for nix-daemon to run as root other than doing build sandboxing (and selectively violating that sandbox) and running findRuntimeRoots?
rauno has joined #nixos
<infinisil>
Nix, as our true lord and saviour shall be granted the power of root as the Gods would have it not any other way
hakujin has quit [Ping timeout: 276 seconds]
<rauno>
:)
<bind>
lejonet: thank you for the explanations
lonokhov has joined #nixos
coot has quit [Quit: coot]
<bind>
lejonet: so I could basically also write a small rust (c or maybe golang?) program to do that stuff too?
<lejonet>
bind: in theory, yes :)
jb55 has quit [Ping timeout: 268 seconds]
<lejonet>
goddamnit nix, make up your mind, first you whine that its a set but you wanted a list, and now you want a set, but its a list!
<bind>
lejonet: are your configurations online somewhere?
<bind>
lejonet: btw
<lejonet>
bind: nope, atleast not that I know off... xD
<bind>
lejonet: oke ... pity :D
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
coot has joined #nixos
<lejonet>
I could give you my /init script if you want, nothing sensitive there :) (oh and a ls -R of the initramfs directory)
<bind>
coming from gentoo (i guess) you find nixos appealing (not talking about the nix language....)?
<lejonet>
Yeah, I really like the parts of nixos and nix I've come across, except that nix is being pesky at me :P (due to my inexperience with it)
<lejonet>
The whole declarative language to conf your entire system except for user-data is awesome
<lejonet>
makes it dead simple to keep backups, just backup /etc/nixos and whatever user-data you want to save and you can restore your system wherever
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rycee pushed 1 new commit to master: https://git.io/vNVst
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master cf3ada0 Ertugrul Söylemez: perlPackages.MailPOP3Client: init at 2.19
<lejonet>
infinisil: rg that you used to search, is that the silver searcher, or the improved version of that, which I always forget the name of?
peacememories has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
ripgrep
<Yaniel>
<3 rg
<Yaniel>
ag is silver archer
<lejonet>
I think nixops is whining at my warnings attribute, but dunno how its supposed to be implemented, so code diving again :P
<bind>
thank you lejonet
<lejonet>
bind: np ^^
<bind>
is there some place where the nix configuration for a given generation is stored?
<bind>
I mean that you can look at the configuration that resulted in the generation ... (without using git)
<bind>
curiosity strikes :D
<lejonet>
Been wondering that myself
civodul has joined #nixos
hamishmack has joined #nixos
<infinisil>
nope
<bind>
buggers... forgot to list nmtui in packages ... and now I have got no wifi grrr
<infinisil>
you'll have to back up the nix files yourself
<bind>
back to the live usb
<infinisil>
nix doesn't know anything about the nix files after evaluation
<bind>
thank you infinisil, is the evaluation result stored somewhere then?
<infinisil>
well the derivations are the result
<lejonet>
infinisil: do you know why the implementation of assertions differs so greatly from warnings?
<infinisil>
lejonet: just looked that up myself, i have no idea
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 54ead73 John Ericson: generic-builder: Make GHC a proper dependency...
<NixOS_GitHub>
nixpkgs/master 5a6b6b4 John Ericson: ghc: Normalize derivations a bit before cross...
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 17 new commits to master: https://git.io/vNVGo
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master 0eb3aca John Ericson: ghc: `echo -n` mk/build.mk to avoid extra line
<lejonet>
Assertions is a list of sets, whereas warnings seems to only support a single warning :S
<lejonet>
Seems really odd honestly
<lejonet>
Tho I'm happy its not the other way around, because its more important to support several assertions than warnings
fxkr has quit [Ping timeout: 268 seconds]
<lejonet>
uhm okay, its whining at something else arbitrarily and pointing to where in the /etc/nixos/nixpkgs/lib/attrsets.nix it whines at O.o
jensens has joined #nixos
<lejonet>
Oh, trace seems to hint that its mapAttrsToList that is called by a type that is whining O.o, would've been very helpful for it to say which... xD
rauno3 has quit [Remote host closed the connection]
<infinisil>
lejonet: oh, it actually kinda makes sense to have different implementations for assertions and warnings: assertions /need/ to check some property, so it makes sense for them to be a (predicate, reason) tuple, a warning however can be present without a condition, therefore it only being a string
<lejonet>
infinisil: Yeah, I can buy that explanation :)
<lejonet>
Would be nice tho if you could trigger several warnings without having to do some type of map or so :P
<infinisil>
what do you mean? warnings is a list
rauno3 has joined #nixos
<lejonet>
infinisil: not from what I can find with some services atleast :P, look at gogs.nix line 265 for example
rauno3 has quit [Read error: Connection reset by peer]
phdoerfler has quit [Quit: Leaving.]
jb55 has quit [Ping timeout: 246 seconds]
<infinisil>
lejonet: apparently setting it to a non-list automatically makes it a single element list (just found that out too)
fxkr has joined #nixos
<lejonet>
infinisil: interesting lol
<infinisil>
but from modules/misc/assertions.nix you can see it's a list
<infinisil>
(the definition of warnings is there)
<infinisil>
and some modules use a list of strings too
<lejonet>
oh well, I only need to output a warning for one thing anyway
<lejonet>
so an optional (stuff) ''Danger! Will robbinsson!''; works
makefu has quit [Ping timeout: 268 seconds]
stepcut has quit [Remote host closed the connection]
<bind>
nix-collect-garbage! :D
<infinisil>
yeah
phdoerfler has joined #nixos
babyflakes has quit []
<lejonet>
raaawr, why can't nix give me which option its whining at, I can't find anything wrong with either of the options I have which allows an attrset
earldouglas has quit [Quit: leaving]
babyflakes has joined #nixos
<phdoerfler>
infinisil: is there a workaround you know of?
<infinisil>
lejonet: your current nixos module declaration is?
<phdoerfler>
infinisil: for the rustc thing
<lejonet>
infinisil: Hmm?
rauno3 has joined #nixos
<infinisil>
phdoerfler: nope, not even using rust (yet?)
<infinisil>
lejonet: i could help you debug if you showed the current nix file
<lejonet>
infinisil: its the same one as before, let me fetch the link again
<bind>
someone in here by accident that is using xmonad or some other tiling/minimal wm with nixos?
rauno3 has quit [Read error: Connection reset by peer]
rauno3 has joined #nixos
<infinisil>
bind: am using xmonad
ckauhaus has quit [Remote host closed the connection]
<bind>
great!
<lejonet>
infinisil: so it has to be an entire big set, not several sets in a list, that gets merged?
jensens has joined #nixos
hakujin has quit [Ping timeout: 265 seconds]
<infinisil>
no, a list of sets, gets merged into 1 set
<infinisil>
but if you have mkMerge [ ... ] ++ map ...
<bind>
infinisil: do you use it in conjunction with some desktopManager and displayManager? or standalone with startx?
<bind>
if you can name that standalone haha
<infinisil>
lejonet: then it results in {} ++ map ... -> {} ++ [ ... ] which doesn't make sense, which is why you got the error about a list being expected
ckauhaus has joined #nixos
<earldouglas>
Is there a vi package in nixpkgs that doesn't just symlink to vim?
<lejonet>
oh, the mkMerge is evaluated before the ++, that does make sense
<averell>
i use xmonad with displaymanager.lightdm.enable = true. works fine
<infinisil>
bind: i just use xmonad with a displayManager (sddm)
makefu has joined #nixos
<lejonet>
infinisil: will that make mkMerge put together the list before making it into one attr set or?
<infinisil>
earldouglas: nvi
<bind>
averell and infinisil thank you, and other sofware you just install piece by piece what you want?
<bind>
maybe a dumb question...
<bind>
:D
<lejonet>
I guess as long as it doesn't except stuff in hardcoded FHS paths, it should work to do that
<bind>
it was so long ago I used xmonad for the last time
<infinisil>
lejonet: hard to say "before" with functional lazy code, it just evaluates what it needs whenever it needs it
<lejonet>
expect*
<averell>
i have most of my installed software in the base configuration, only some more rapidly changing stuff (wrappers) in the user install
<infinisil>
bind: yeah, i just install whatever i need
fyuuri has joined #nixos
<lejonet>
infinisil: but basically, for it to evaluate the way I want/need, I need to ensure that the list is finalized before handed to mkMerge, right?
<infinisil>
lejonet: finalized?
<averell>
i use taffybar for systray and stuff, but i don't think it's very popular
bpa has joined #nixos
rauno3 has quit [Ping timeout: 246 seconds]
<lejonet>
infinisil: finalized as in all the optional parts have been added to it
ckauhaus has quit [Remote host closed the connection]
<lejonet>
i.e. the ++ have to have been evaluated on the list already
ckauhaus has joined #nixos
<bind>
infinisil or averell and does sound work out of the box? (can remember back then it was a major struggle to get things working well together in debian for me)
<infinisil>
lejonet: you just tell what the list is, use mkIf where needed, and it'll work in the end
<averell>
sound is easy, bluetooth is a bit of a pain still
<infinisil>
lejonet: I might not really get your question
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg merged master into unstable-aarch64: https://git.io/vNVcb
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
infinisil: wouldn't be the first time I fail to explain myself, today :P
<bind>
taffybar .. with built in tray nice :D
<bind>
averell: bluetooth is a pain?
<infinisil>
bind: i just have hardware.pulseaudio.enable = true; which works no problem
<averell>
yes, a bit flaky compared to non-linux
<lejonet>
infinisil: basically, I was thinking of making the list in a separate variable, then mkMerge that variable, because then the entire list, appends and all, should've been evaluated before being handed to mkMerge
krey has joined #nixos
<infinisil>
lejonet: you seem to still be thinking imperatively
<lejonet>
infinisil: ... yep :P
<infinisil>
in functional code, `let x = 1; in f x` is exactly the same as `f 1`
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] leenaars opened pull request #34161: pythonPackages update: asn1ate and asn2quickder (master...asn2quickder) https://git.io/vNVCq
NixOS_GitHub has left #nixos [#nixos]
<krey>
o/ guys, does anyone know how I can debug a package being built from source even though it should be pulled form the binary cache?
<bind>
Will just try it out then ... thank you for the input averell and infinisil (again) :)
<lejonet>
infinisil: so what you're saying is that I need the paranthesis horror... :(
<infinisil>
lejonet: ah, you can do a `let in` if you want, to have less of a mess, but it doesn't make a difference to nix, at all
<lejonet>
infinisil: so let list = [ stuffff ] ++ map ++ map ++ map; in mkMerge list; ?
<infinisil>
yeah
<lejonet>
This is one of the hardest parts to understand when your head just isn't into functional mode (dual-meaning!), how the heck to add sequential dependencies in code... :P
<infinisil>
krey: You could see if it failed on hydra and the logs of it (because if it failed to build on hydra it won't be in the cache): https://hydra.nixos.org/jobset/nixpkgs/trunk
<infinisil>
lejonet: yeah, in functional programming there really isn't any sequential stuff, nothing happens before or after anything, actually nothing "happens" at all. It's all just a declaration
<infinisil>
you just tell it how it is
sethetter has left #nixos [#nixos]
<krey>
infinisil: i see a lot of green ticks (this is for libreoffice), so I guess it's not hydra's fault
<infinisil>
and nix can't do anything like printing to stdout which would require some sequence anyways
shabius has quit [Quit: Leaving]
<infinisil>
krey: oh libreoffice, I've seen some people complain about that too
<MichaelRaskin>
Well, a failing LO build has been seen (a segfault)
<MichaelRaskin>
Well, maybe a (trivial) LO rebuild will succeed now
* bind
will shut up again for now :P
mahalel_ has joined #nixos
ckauhaus_ has joined #nixos
xcmw has joined #nixos
<infinisil>
bind: I'd love some xmonad similar thing for wayland (there is actually something like that, don't know the state of it though: https://github.com/Ongy/waymonad)
<kier>
is there some manner in which I can politely bump a PR? Specifically, https://github.com/NixOS/nixpkgs/pull/33287 is tested and should be OK to merge, but it seems to have stagnated
asuryawa_ has quit [Ping timeout: 255 seconds]
ckauhaus has quit [Ping timeout: 268 seconds]
<dtz>
idk I just say "ping! :)" and no one's been too grumpy at me AFAICT... :innocent:
shabius has quit [Client Quit]
<infinisil>
kier: (you just did by asking here :P)
<dtz>
clarifying/reiterating why it's important and what testing has been done (especially if it's beyond what's mentioned in the original PR) helps
<dtz>
basically as long as you're polite and (vague) "mean well" you'll be fine
<dtz>
esp rem most folks are a bit overwhelmed, limited on time, but doing best they can :)
<bind>
infinisil: I know xmonad isn't moving into wayland direction and actually was wondering about that too... would be nice to have something extensible (in a safe-enough programming language) ... there is way cooler, which you can extend in lua or via dbus I guess but from "shallow" searching in the last few days I didn't see that much movement
<infinisil>
Damn, 618 PR's
<krey>
MichaelRaskin: there seem to be two successful builds after that?
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 2 new commits to staging: https://git.io/vNV8Z
<NixOS_GitHub>
nixpkgs/staging 4e5a4a9 John Ericson: Merge pull request #33478 from dtzWill/fix/cross-bc...
<NixOS_GitHub>
nixpkgs/staging 8b21391 Ben Gamari: bc: fixes for cross-compilation
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
which is a list of lists, but mkMerge wants a list of attrsets
<infinisil>
that's why I used an additional mkMerge in my example
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] primeos pushed 1 new commit to master: https://git.io/vNV8B
<NixOS_GitHub>
nixpkgs/master a03b2b9 Michael Weiss: tdesktop: Fix the execution in a pure environment...
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
lejonet: you can wrap the maps with mkMerge and it should work
<lejonet>
The interesting part about that is that :p map (mon: { systemd.services."ceph-mon@${mon}".enable = true; } ) daemons in nix-repl, with daemons = [ "0" "1" ]; gave me this output: [ { systemd = { services = { "ceph-mon@0" = { enable = true; }; }; }; } { systemd = { services = { "ceph-mon@1" = { enable = true; }; }; }; } ] which I assumed was a list of attrs
<infinisil>
it is a list of attrs
<infinisil>
but optional true (map ...) isn't
<lejonet>
I see
<lejonet>
blargh xD
<krey>
ah, I can't even find libreoffice. does nix-env -qaP libreoffice-fresh give you guys anything?
<infinisil>
krey: it does not, never used this nix-env search much though, I'm always doing `nix-env -qaP | fzf` which when typing libreoffice, lists libreoffice-fresh
<infinisil>
(nixos.libreoffice-fresh)
<lejonet>
infinisil: nix-repl confirms your theory
<krey>
infinisil: not a bad idea. are you able to install it `nix-env -i libreoffice-fresh --dry-run`?
<infinisil>
krey: errors, but nix-env -iA nixos.libreoffice-fresh --dry-run works
<bind>
nn
<krey>
infinisil: huh, interesting
<krey>
infinisil: thanks
<infinisil>
:)
<lejonet>
infinisil: what am I doing wrong in nix-repl if I get "error: attempt to call something which is not a function but a set" when trying to use lib.mkMerge in it? I'm fairly certain I'm giving it the value wrongly xD
<lejonet>
screw this, mkIf seems to be doing this sanely as opposed to optional... :P
mkoenig has quit [Remote host closed the connection]
<infinisil>
lejonet: how are you using mkMerge?
<lejonet>
infinisil: like a tard probably, I honestly don't know how I should give it the output from map without it becoming even more scrambled...
<lejonet>
both lib.mkMerge lib.optional true (map (mon: { systemd.services."ceph-mon@${mon}".enable = true; }) daemons) and lib.optional true (lib.mkMerge map (mon: { systemd.services."ceph-mon@${mon}".enable = true; } ) daemons) gives me that error
mkoenig has joined #nixos
zzamboni has quit [Quit: Leaving.]
<infinisil>
lib.mkMerge (map (...
<infinisil>
needs brackets
<lejonet>
lib.mkMerge [(map (... ?
<infinisil>
no, because map outputs a list itself
michaelpj_ has quit [Ping timeout: 240 seconds]
jrolfs has quit [Ping timeout: 276 seconds]
<lejonet>
so I need to wrap () around the map call?
<infinisil>
yeah
<lejonet>
That gives me "[ error: attempt to call something which is not a function but a set, at (string):1:20" instead :P
bpa has quit [Read error: Connection reset by peer]
dan_b has joined #nixos
<krey>
infinisil: I swear I'm not a total noob, but how can I uninstall it now? `nix-env -e nixos.libreoffice-fresh` doesn't do anyting :/
<infinisil>
because otherwise it will pass the arguments "map" and "(mon: ..." to mkMerge, which only takes 1 argument of a list of attributes
bpa has joined #nixos
<lejonet>
infinisil: fair enough
quam6246 has joined #nixos
<bind>
one more question ... about packages ... it it possible to "track" some channel for a certain package (e.g. docker) while staying on release for the rest? or would it be wiser to just use master ?
<infinisil>
krey: Now I'll give you the master tip, you'll use this to uninstall everything from now on because it's just so good: nix-env -q | fzf | xargs nix-env -e
<infinisil>
lejonet: your current nix expression is?
<lejonet>
I might've missed one ) now that I look at it
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg force-pushed unstable-aarch64 from 617e9ba to 9848df4: https://git.io/vSSlm
<NixOS_GitHub>
nixpkgs/unstable-aarch64 9848df4 Tuomas Tynkkynen: nixos/release.nix: Introduce callSubTestsOnTheseSystems
NixOS_GitHub has left #nixos [#nixos]
<lejonet>
infinisil: found the error, forgot that daemon was part of input to map, so had placed the ) wrongly
<infinisil>
bind: you could just do something like `(import (builtins.fetchTarball "github.com/NixOS/nixpkgs-channels/archive/17.09.tar.gz") {}).your-package`
strobelight has quit [Quit: gtg]
<infinisil>
bind: which would potentially fetch it every hour, or you could probably do something with a separate channel with nix-channel, call it foo and use `(import <foo> {}).your-package` (not sure though, never did this)
<krey>
infinisil: what is this sorcery? :O
<krey>
infinisil: thank you!
<bind>
infinisil: so ... that import thing and then just add docker to the packages list?
quam6246 has quit [Ping timeout: 248 seconds]
<infinisil>
bind: you'll want to use the import thing for your docker package, the import thing actually imports the specific nixpkgs version, if you use docker from there you'll get the specific nixpkgs docker
<kier>
I wonder if it's possible to override a module to a version from a different channel in the same way
<bind>
infinisil: you mean that that does not go into configuration.nix?
<infinisil>
bind: e.g. let stable = import ...; in at the beginning of your file, then environment.systemPackages = with pkgs; [ ... stable.docker ];
<bind>
infinisil: I see
<kier>
I vaguely remember seeing an issue that seemed related a couple weeks ago
hiratara has quit [Ping timeout: 265 seconds]
<bind>
is tracking master a good idea? or should I rather stay on release? :)
dannyg has joined #nixos
<bind>
how often do things break?
<infinisil>
kier: the best way to do this currently is to check out the different version of the module in a nixpkgs checkout
<LnL>
use a tested channel for your system
<bind>
okay :)
<MichaelRaskin>
Yes, definitely (said me, who uses master for the system)
<LnL>
but for user packages you can do whatever you want
<infinisil>
bind: specifically all nixos-* channels are fine
<ixxie>
which editor has the better support for a Nix-based Haskell development workflow?
<bind>
LnL: thank you :)
<lejonet>
infinisil: well the mkMerge parts seems to have worked, now its whining that an attribute is missing, which is an option that has a default value
<ixxie>
I am currently using emacs but considering to switch to vim/neovim
<lejonet>
infinisil: I'm guessing that is due to not having rec, so that it cannot fetch the value of a config option?
<infinisil>
lejonet: what attribute?
zzamboni has joined #nixos
simukis has quit [Ping timeout: 240 seconds]
<lejonet>
infinisil: clusterName
dannyg has quit [Quit: dannyg]
kalebo has joined #nixos
<lejonet>
Its referred in the Environment of each of the services :/
<infinisil>
lejonet: cfg.global.clusterName
<infinisil>
instead of cfg.clusterName
<lejonet>
ah right, forgot that I renamed it -.- man I'm getting tired
<lejonet>
xD
<infinisil>
:P
krey has quit [Ping timeout: 264 seconds]
kalebo has quit [Client Quit]
ckauhaus_ has quit []
<lejonet>
infinisil: yay, now its whining at my user conf, not the service xD
<infinisil>
writing a nixos module is a pain, i have to admit. Nix's errors are horrible, probably one of the worst ones of any programming language, and there's no easy way to debug it
<lejonet>
after today, I fully agree with you there
<infinisil>
you can post the error here
<lejonet>
infinisil: isn't types.commas supposed to mean types.listOf types.str?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rasendubi opened pull request #34167: screenkey: init at 0.9 (master...screenkey) https://git.io/vNVEF
NixOS_GitHub has left #nixos [#nixos]
iLikeSALAD has joined #nixos
<lejonet>
infinisil: I'm giving up for today, biiig thanks for all the help today
bfrog has joined #nixos
mizu_no__ has joined #nixos
<infinisil>
lejonet: let's you assign a list of strings, but will result in a string that concatenates everything with a comma
jb55 has joined #nixos
<infinisil>
ah wait, not a list
<lejonet>
infinisil: because I had that and gave it [ "something" ] and it said "is not null or a string" which kindof threw me off, I wanted to assign a list, that gets converted into a comma separated list
MP2E has joined #nixos
<iLikeSALAD>
Yo, i got aarch64 nixos on a RPi 3, hdmi output works with my tv... BUT on my monitor its a whole different story! it gets to ‘starting kernel’ then monitor screen goes black but can still detect an input since it does not turn off. Any ideas why this happens?
<infinisil>
lejonet: probably best to use listof str then and manuall concatenate it
<lejonet>
infinisil: yeah, going to be interesting to see what toINI does with that :P
<infinisil>
alright well glad i could help you lejonet :)
<infinisil>
I should be going to bed now
iLikeSALAD has left #nixos ["Be back later..."]
<alexteves>
how do I try nix 1.12? I switched to unstable-small, did "nixos-rebuild switch --upgrade" & "nix-env -u nix"
<alexteves>
am still on 1.11.16
<infinisil>
alexteves: nixUnstable
<alexteves>
nix-env -i that?
iLikeSALAD has joined #nixos
<infinisil>
alexteves: e.g. nix-shell -p nixUnstable or nix-env -iA nixos.nixUnstable
<alexteves>
installing ‘nix-1.12pre5788_e3013543’
<alexteves>
nice
ylwghst has quit [Ping timeout: 268 seconds]
<alp>
you can also specify nix.package = nixUnstable in configuration.nix if you want to use it everywhere IIRC
<lejonet>
infinisil: just a quick question before I disappear for real xD "attempt to call something which is not a function but a set" at line 35, char 28, which is "type = with types; nullOr listOf str;" and char 28 is the n of nullOr, did I screw up the with types; ?
<infinisil>
which makes sense if you think about it
<infinisil>
nullOr doesn't take 2 arguments
<lejonet>
yeah
<lejonet>
I just haven't understod when its implicit and when its explicit... :P
ryanartecona has quit [Quit: ryanartecona]
knupfer has quit [Ping timeout: 246 seconds]
knupfer1 is now known as knupfer
coot has quit [Quit: coot]
iLikeSALAD has quit [Read error: Connection reset by peer]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34168: aview: init at 1.3.0rc1 (master...feature/aview) https://git.io/vNVzb
NixOS_GitHub has left #nixos [#nixos]
threepointone has joined #nixos
<infinisil>
oh, actually, I packaged the terminal mandelbrot renderer a couple months ago and never made a PR for it
mog has left #nixos ["Leaving"]
mog has joined #nixos
mizu_no__ has quit [Quit: Computer has gone to sleep.]
<threepointone>
So, I installed some packages as root and then as a normal user. Now, the output of "nix-env -q" is different for root and the user but I'm able to use packages installed by root as the normal user and vice versa. I though that "Packages installed using nix-env are only available to the given user" but that's not what's happening here is it?
szicari has quit [Quit: szicari]
<mog>
threepointone, thats true for users who arent root
<mog>
roots packages leak
<mog>
otherwise you could only install packages globably via configuration.nix
knupfer has quit [Ping timeout: 264 seconds]
quam6246 has joined #nixos
<threepointone>
mog: Is there any reason why I would install packages as $ nix-env and not as # nix-env ?
<mog>
if you wanted a package to only be accessible to a user
<mog>
i tend to install things in my user $ nix-env
<mog>
and then if i decide to keep it i add it to my configuration.nix later
<threepointone>
mog: But the packages that I installed as $ nix-env are accessible to root and the packages I installed via # nix-env are accessible to the user. Does the statement "only available to the given user" apply when we have multiple normal users?
<mog>
no packages installed in $ user arent accessible to root
hiratara has joined #nixos
quam6246 has quit [Ping timeout: 246 seconds]
<mog>
unless # root also installed the same name
<threepointone>
mog: In my VM, they are.
<mog>
nope
xcmw has quit [Ping timeout: 256 seconds]
<mog>
or thats not how its worked on any nix system i have used
jrolfs has joined #nixos
<threepointone>
mog: If I installed a package using $ nix-env, I installed it in my $, right?
<mog>
yes
<threepointone>
So, you're saying that it shouldn't be accessible to root?
<mog>
but just your users environment , not even other $ users
<mog>
correct
<threepointone>
Right now, $ nix-env -q shows that I have htop and neofetch. However, when I switch to the root user, I can't see those packages as installed when I type # nix-env -q. However, I'm able to use htop just fine as root.
<infinisil>
threepointone: what does `which htop` output?
<threepointone>
infinisil: as root or as the user?
<infinisil>
both preferably
<infinisil>
or `type htop` if which doesn't work
<threepointone>
infinisil: both of them output "/home/user1/.nix-profile/bin/htop
<mog>
how did you become root?
<threepointone>
mog: using su
<mog>
su - ?
<threepointone>
mog: no, just su
<infinisil>
Ohh, well su itslef just keeps env vars
<mog>
correct
<mog>
without the dash you arent resetting the env
<infinisil>
so you still have $PATH from your normal user
<mog>
so its inheritted
<threepointone>
infinisil: and mog: Yeah, I logged out from the normal user and logged in as root and I'm not able to see htop anymore. That did it.
<threepointone>
But is this really expected? I mean, isn't there supposed to some sort of security barrier so that different users can't access each others packages? Is it just about gaining access to the correct $PATH value?
<mog>
threepointone, thats how su has always worked
<threepointone>
Sorry if I'm asking stupid questions.
<infinisil>
threepointone: also, nothing is preventing anybody from just executing stuff from store directly
<infinisil>
threepointone: you can execute anything from any other user, just go find it in /nix/store and execute the binary directly
<infinisil>
(you just don't know which user uses what, you jsut know it's there)
<catern>
man! nix is so freeing
<catern>
I was thinking about how I was going to deploy my fancy Nix setup, and I thought, maybe I would want to use this certain program which requires Python 3.6
<catern>
and I briefly worried, "oh no, then I have to painfully deploy Python 3.6 too"
<catern>
But no, no problem - just use Nix :)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Infinisil opened pull request #34169: almonds: init at 2015-12-27 (master...add/almonds) https://git.io/vNVaQ
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
catern: \o/
<infinisil>
Also, here is my PR to add almonds, a pretty neat mandelbrot set explorer for the terminal, if anybody wants to try it out :)
lonokhov has quit [Ping timeout: 260 seconds]
stphrolland has quit [Ping timeout: 260 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 5 new commits to master: https://git.io/vNVVV
<NixOS_GitHub>
nixpkgs/master 728a132 Mitsuhiro Nakamura: cataclysm-dda{,-git}: build on Darwin
<NixOS_GitHub>
nixpkgs/master 7287a9e Mitsuhiro Nakamura: cataclysm-dda{,-git}: create app launcher on Darwin
NixOS_GitHub has left #nixos [#nixos]
elurin has quit [Quit: ERC (IRC client for Emacs 25.3.1)]
erictapen has quit [Remote host closed the connection]
ryanartecona has joined #nixos
<threepointone>
infinisil: Is there a reason why the binaries in /nix/store are executable for everybody? What if we set the permissions to 500 or maybe 550 instead of 555 and change the ownership to the user who installed the package? To prevent reinstalling the same package if another user installs it, we can add the user to the group of user already allowed to access the package during the installation.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34170: flex: fix for cross (staging...fix/flex-cross) https://git.io/vNVwZ
NixOS_GitHub has left #nixos [#nixos]
ryanartecona has quit [Client Quit]
erictapen has joined #nixos
<infinisil>
threepointone: there's a whole lotta debate about permissions for /nix/store which I don't want to get into right now (and i haven't really thought much about it), probably there's a couple nix issues floating around though
xcmw has joined #nixos
<__red__>
Quick Question: I made some changes to a .nix file, committed it to my github. Saw the error of my ways, revertted it. Wandered the universe for a few months... just fetched upstream, merged with upstream and pushed to my github repo so I'm current BUT... my github.com/ says ythat I'm ahead of nixos/nixpkgs master by 4 commits
<__red__>
the 4 commits should be a NO OP
alexteves has quit [Ping timeout: 248 seconds]
<__red__>
any advice on how I can remove / deal with them so they don't appear in any future PR I try to put together? please
<infinisil>
__red__: probably more of a #git question, but you can force revert your branch to upstream with `git reset --hard upstream/master`
<__red__>
infinisil: thank you - that'll do nicely
<__red__>
I remember I used to just delete the repo and fork it again but then learned that lost history of association
<joko>
__red__: or do git rebase upstream/master master
<infinisil>
__red__: (note the --hard in there, it will not persist history if you have it in master, so be sure to keep the important commits on another branch or so)
<__red__>
well, all the cjanghes from my previous PR live in nixos master now, so a hard reset I think is probably the cleanest outcome?
<joko>
__red__: then yes
<__red__>
I have no commits I want to keep that have not already been merged in by you guys
<infinisil>
i suggest you to always use branches for doing stuff, keeps it a lot cleaner
<infinisil>
ALright, now I got a question too: What do I do for packages that don't have a license?
<NixOS_GitHub>
[nixpkgs] deepfire opened pull request #34172: ghc841: update to 8.4.1-alpha2 (master...x-ghc-8.4-master) https://git.io/vNVrC
jb55 has quit [Ping timeout: 260 seconds]
<__red__>
I just need the commit and revert to disappear I guess or it's going to end up being picked in every PR I submit
<__red__>
I guess I could also maybe rename this branch. nuke it, create a new master from upstream
* __red__
shrugs
<infinisil>
(i will answer git questions in #git)
<__red__>
I'll try #git
<__red__>
thanks
<__red__>
omw
blym has quit [Read error: Connection reset by peer]
mizu_no_oto has quit [Client Quit]
allanon_ has joined #nixos
<allanon_>
Hi there. I have an unfortunate newb question: I am running the package manager on High Sierra and have run into bug #1451 (dyld issue). --version returns 1.11.6, so it looks like I need upgrading. --upgrade returns silently, however, and my --version remains at 1.11.6. So ... how do I upgrade nix itself?
quam6246 has joined #nixos
jsgrant[m] has joined #nixos
<infinisil>
allanon_: I'll be going to bed now, but you might want to ask in #nix-darwin as well which is a bit more specific (but here is totally fine too)
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34173: misc minor cross fixups from Ben (staging...fix/cross-misc-ben-1) https://git.io/vNVoF
NixOS_GitHub has left #nixos [#nixos]
<aminechikhaoui>
allanon_: what channel are you using ? you might try nix-env -i nix which should update nix to the latest in your channel
ryanartecona has joined #nixos
mbrgm has joined #nixos
<threepointone>
infinisil: Is this https://github.com/NixOS/nix/issues/8 one of the open issues that you were talking about? Oh man, this is open since 2012 and Eelco himself opened it. Is there a specific reason why implementing permissions and ACLs haven't been done in the /nix/store yet?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 1 new commit to release-17.09: https://git.io/vNVK2
<NixOS_GitHub>
nixpkgs/release-17.09 4ea23e7 John Wiegley: nss: Omit an extraneous definition; fix other problems on Darwin...
NixOS_GitHub has left #nixos [#nixos]
Itkovian has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tg has joined #nixos
<allanon_>
aminechikhaoui: I ran nix-env -i nix. Now --version yields "dyld: Library not loaded". So it got worse heh.
<samueldr>
the long awaited break from now until the end of christmas 2018?
<petersjt014[m]>
sooo I tried installing nixos on windows with wsl
<jefflovejapan>
Hi all, I was hoping someone might be able to help explain what's going on with our build setup. We're running make against the GNUmakefile located here: https://pdos.csail.mit.edu/6.828/2017/jos.git
<petersjt014[m]>
and besides being slow, it works surprisingly well
<samueldr>
full nixos?
<petersjt014[m]>
just nix, sorry
<samueldr>
ah
<MichaelRaskin>
Ah
<petersjt014[m]>
I have found only one problem
<Sonarpulse>
dtz: haha ok!
<jefflovejapan>
and getting errors claiming "undefined reference to __stack_chk_fail_local"
<Sonarpulse>
btw you know about #nix-dev ?
<petersjt014[m]>
the commands (nix-env etc) and commands for pks installed with nix-env get unaliased sometimes
<petersjt014[m]>
anybody else hadthat happen?
<jefflovejapan>
We think that GCC is getting invoked with "-fno-stack-protector", but we're not sure
xcmw has quit [Ping timeout: 264 seconds]
<petersjt014[m]>
rerunning the install script works, but that seems like overkill
quam6246 has joined #nixos
<jefflovejapan>
the reason we think that is because of this line in the make file: CFLAGS += $(shell $(CC)) -fno-stack-protector -E -x c /dev/null >/dev/null 2>&1 && echo -fno-stack-protector)
allanon_ has quit [Quit: Page closed]
<jefflovejapan>
we've tried adding the -lssp flag to LDFLAGS but still getting similar errors