<grantwu>
gchristensen: Do you know what the reason for splitting out Kerberos support is anyways
<gchristensen>
I have no idea, but maybe it makes the closure much bigger?
ottidmes has quit [(Ping timeout: 268 seconds)]
<gchristensen>
garbas: if you find issues, please open bug reports :)
<gchristensen>
grantwu: maybe open an issue about merging the two + an email to the mailing list?
<gchristensen>
grantwu: it was added in 2c089337e7a4260cec5b83f98754e5dc8445a0f4 in 2013 and I'm not sure it was made separate for a reason or because the contributor didn't feel comfortable choosing to add it to openssh by defeault
<grantwu>
Hrm
<grantwu>
I will probably open an issue then
phreedom has joined #nixos
<gchristensen>
maybe tonight I'll try to backport ff57
<gchristensen>
it'd be good (read: very important) to get that backported and I'm not sure anyone has really put the time in to getting the right rust in to 17.09 for it
thblt has quit [(Remote host closed the connection)]
thblt has joined #nixos
ssmike1 has joined #nixos
ssmike has quit [(Ping timeout: 258 seconds)]
ssmike1 is now known as ssmike
vandenoever has quit [(Ping timeout: 248 seconds)]
thblt has quit [(Ping timeout: 260 seconds)]
erictapen has joined #nixos
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
<antonv>
samueldr: I do use sapon, but this doesn't extend the /nix/store
<antonv>
under the installer image the /nix/store is a unionfs of two directories
<samueldr>
hm, I thought tmpfs would have automatically used swap when available, sorry
<antonv>
/nix/ro-store - read only [artfrom the installer image
<antonv>
and /nix/rw-store - a tempfs, of 1GB site
<antonv>
1GB is not enough
<antonv>
siZe
<antonv>
I would like to reconfigure the unionfs /nix/store/ by adding to it a free space from my flash drive
<antonv>
but I can't unmount /nix/store
<samueldr>
mount -o remount,size=____ /proper/mount/point where _____ can be something like 4G
griff_ has joined #nixos
dan_b has quit [(Ping timeout: 248 seconds)]
boxofrox has quit [(Ping timeout: 268 seconds)]
Oida has quit [(Ping timeout: 258 seconds)]
<avn>
(semi-offtopic) Folks, does anyone here have some expirience with modern (pci-e) scsi controllers/cables etc? (sorry for repeat)
Oida has joined #nixos
astronav_ has joined #nixos
astronavt has quit [(Ping timeout: 250 seconds)]
aarand has quit [(Ping timeout: 250 seconds)]
astronav_ is now known as astronavt
boxofrox has joined #nixos
aarand has joined #nixos
infinisil has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
griff_ has quit [(Quit: griff_)]
infinisil has joined #nixos
katyucha_ has quit [(Ping timeout: 240 seconds)]
nuncanada has joined #nixos
<antonv>
samueldr: unionfs is a fuse filesystem, unfortunately it doesn't support -o remount
grantwu has left #nixos ["WeeChat 2.0"]
<samueldr>
I was thinking about the tmpfs that's under the unionfs
<samueldr>
(if I understood your problem correctly)
phdoerfler has joined #nixos
<antonv>
your understanding seems correct - yes, there is a tempfs as a part of unionfs
<antonv>
will unionfs handle correctly change of the underlying fs - one of the doubts I have
<clever>
it will easily handle the tmpfs changing size
<clever>
and you can add a swap file on anything to handle that new size
<antonv>
BTW, in internet I see program unionctl which should be able to --add new parts to unionfs
<antonv>
but the nixos doesn't have it
<avn>
antonv: is only one way to know -- if it support remount or not ;)
<avn>
s/remount/change size/
<avn>
but I believe, that unionfs-fuse doesn't care about underlying size at all, as long as files fit into it
<clever>
yep
<samueldr>
oh, I love cached failures, stops me from waiting for a rebuild when I didn't really change anything :)
grantwu has joined #nixos
<grantwu>
gchristensen: Wait, why would rustc depend on OpenSSH?
<grantwu>
ah, it depends on Git.
<clever>
grantwu: something to do with git clone i think, the whole "rewrite it in rust" thing went a bit far :P
<grantwu>
sigh.
<grantwu>
clever: rewrite what in Rust?
<clever>
git?, or did it just depend on normal git
<grantwu>
guh, bit rustc doesn't really _link_ against git, right?
<grantwu>
Does it really make sense that I have to rebuild rustc because git changed
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vbR1L
<NixOS_GitHub>
nixpkgs/master 8a7b2f8 Shea Levy: insert_dylib: Set platforms attribute
NixOS_GitHub has left #nixos []
<disasm>
hey folks, setting up my new nixos router :) Is it possible to assign multiple interfaces to a vlan? Or do you need to have a bridge setup in that scenario?
<clever>
disasm: i think you would need to bridge them
<grantwu>
I am now building the rustbook -_-
<clever>
6: wan@enp4s2f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default
<clever>
disasm: after configuring the vlan on my router, this wan interface showed up
<clever>
most tools can accept just 'wan' as the name, but 'ip link' shows the master IF it came from
<clever>
and i also did networking.interfaces.enp4s2f0.useDHCP = false; so no un-tagged dhcp queries leak out
<disasm>
clever: ok, so would you do something like network.vlans = { lan = { interface = "enp2s0"; id = "33"; }; }; and then for the bridge, have it have interfaces enp2s1 and enp2s0.33
<clever>
the config i typed above, creates a new wan interface
<clever>
in the example you typed, i believe it will create an interface called "lan"
<disasm>
clever: so does that tag vlan 35 on your enp4s2f0 port?
Arcaelyx_ has joined #nixos
<clever>
yeah
<disasm>
giving it a shot now :)
<clever>
oh, and normal tcpdump gets the packets after they passed thru the networking stack, and the enp4s2f0 interface gets all packets, tagged or not, and the networking stack strips the tags
<clever>
i think wireshark can see the tags though, it uses different flags for capture
<clever>
that caught me off guard when debuging
<clever>
so enp4s2f0 is the result of all vlan's merged and tag-less, and wan is just vlan 35, filtered and tagless
erictapen has quit [(Ping timeout: 248 seconds)]
orivej has quit [(Ping timeout: 260 seconds)]
<disasm>
clever: cool, deployed at least. do you use ipv6 by chance?
<clever>
disasm: yeah, but thats handled in a half manual way
<clever>
which reminds me, i never fixed it after that power outage
<clever>
i'm using a tunnel, no native support from the ISP
<grantwu>
Is there any effort ongoing to minimize dependencies
<pie_>
Lisanna, if pypi2nix complains about wanting gcc is gcc the right package to pass? i tried stdenv but that didnt work
<grantwu>
Like, I do not need the rust book to use ripgrep, yet when I end up rebuilding Rust, I end up recompiling the rust book
<pie_>
hm ggc didnt work either
* pie_
trawls error logs
<pie_>
distutils.errors.DistutilsExecError: command 'gcc' failed with exit status 1
<clever>
pie_: anything above that, like an error from gcc itself?
dbmikus has joined #nixos
<pie_>
oh good point its an error 1 :/ i need to pay better attention
<pie_>
c/_cffi_backend.c:15:17: fatal error: ffi.h: No such file or directory
<pie_>
probably needs cffi
ilja_kuklic has quit [(Ping timeout: 268 seconds)]
<clever>
adding that to the buildInputs should fix it
<clever>
libffi.dev 13,338 r /nix/store/h0qamldzfs1prm6pi6nrr109sd2016m4-libffi-3.2.1-dev/include/ffi.h
<clever>
libffi should do
<pie_>
clever, im using pypi2nnix so hopeflly its just -e cffi, though i find it strange that it wouldbt be discovered in the package deps
<pie_>
ah yeah ok looks like thats consistent...that didnt work
<clever>
try libffi instead
<clever>
the error is caused by cffi trying to include normal ffi.h
<pie_>
oh i see
<pie_>
that would make a lot more sense
<pie_>
ok worked like a charm :) on to different errors :P thanks.
<clever>
pypi2nnix probably only knows about the python side of the deps, and breaks as soon as it relies on a system lib
<pie_>
yeah of ourse
<clever>
the great thing about using nix, is that it doesnt care what language a package is in
<clever>
you just describe the deps for everything, and it works
<pie_>
i love nix, it just hasnt been user friendly enough for me to learn everything as i go :/
<clever>
pie_: have you tried nix-repl yet?
<pie_>
a little yeah
<pie_>
it helped a lot heh
<clever>
yeah
<pie_>
with all the screwery i do id probably have/wanted to break whatever distro i had at this point lol
<pie_>
if nothing else
<pie_>
if NOTHING else
<pie_>
i dont have to deal with ubuntu complaining about the wrong libc version and not being able to hack together rpaths on binaries by hand or making a whole chroot
<clever>
ive made a disk image before, that had both arm and x86-64 binaries in the same rootfs
<clever>
and it was able to boot on both
<clever>
try doing that on ubuntu, lol
<pie_>
i love what nix theoretically lets me do with environments and stuff
<pie_>
thogh its really annoying not being able to hack easy imperative things sometimes :P
<disasm>
from the br0 or the voip interfaces that is
<clever>
disasm: line 27-28 shouldnt be needed, the networking.nat handles it for you
<disasm>
pings fine from enp1s0
<pie_>
clever, on this note, nix would have probably been the perfect thing to try getting my own little distributed cluster working when i was trying ot make images. ill ahve to give that a shot again sometime
<clever>
pie_: that one file, fully configures network booting of nixos, including setting up NAT
<clever>
the original use, was to plug ANY machine into the ethernet jack of my old laptop, and instantly netboot nixos
<pie_>
cool
<clever>
and it has justdoit on that ramdisk
<pie_>
oh neat
<clever>
so i literally type "justdoit" into the console, and it formats the drive with nixos
<clever>
and the netboot server (the old laptop) shares its wifi over the same ethernet port with NAT
<clever>
that was the plan
<clever>
but uefi and nvme gave trouble
<clever>
i had tested it on non-uefi hardware with sata drives
<clever>
disasm: so enp3s0 has 2 isolated lans on seperate vlans?, and enp1s0 is your wan port?
hotfuzz has quit [(Ping timeout: 268 seconds)]
<pie_>
well, back to coding
<clever>
disasm: oh, wait, i see an issue, your using br0 to link the 2 vlans, of the same port, so they basically ceased to be seperate vlans...
<clever>
oh wait, no, there is a 3rd IF in play...
<jeaye>
Is there a nix approach to running services within nix-shell? For example, I want a nix-shell for development which is also running postgres as per some configs. I can see this working with containers or docker, but I wonder if there's a lighter way.
<clever>
disasm: let me grab dot, lol
<disasm>
clever: so enp1s0 is going to go to cable modem, enp2s0 should be untagged lan (it's a member of br0), enp3s0 has a bunch more, but keeping it simplified right now until this works... it should tag 33 for lan and 40 for voip. that then goes to a trunked managed switch port that has computers (lan) and a phone plugged in (voip)
<disasm>
clever: yeah, aside from it being vlan 33, not 30
<clever>
disasm: can you also gist the ip addr of the nixos machine?
<catern>
argh, this is silly
<catern>
don't I just have to delete /nix/var/nix/binary-cache* to clear the binary-cache-cache?
<catern>
is there anything more than that?
<catern>
because that is not sufficient, it seems...
<clever>
disasm: another thing that helps, start pinging 8.8.8.8 from a machine on vlan 33, then `tcpdump -v -i lan_port -n -p`, and see what shows up
<clever>
catern: what path is it not finding, on which cache?
<catern>
I am deleting it, yet, Nix still seems to have negative cache entries!
<catern>
clever: it's my own private nix-serve cache
* samueldr
is guessing 'ofborg isn't feeling right
<catern>
it's the path to a file called "hello" containing "hello world" :)
<clever>
catern: do you know how to query the narinfo files?
<catern>
yes, the file is definitely
<catern>
there
<catern>
I can wget it just fine
<clever>
and have you restarted nix-daemon after adding the cache to nix.conf?
<catern>
yes - let me short circuit this by saying that I've previously successfully got a different file from this cache
<catern>
now I am testing clearing negative cache entries
<catern>
because that has confused me in the past
<disasm>
clever: I'm not even that far yet. Still trying to get pings using -I interface to work right
<clever>
disasm: pings to which machine?
<disasm>
clever: oddly it doesn't even show anything if I tcpdump -i br0 icmp and run ping -I br0 8.8.8.8
<clever>
it has to use ARP to ask the local network who owns 8.8.8.8
<clever>
and nobody owns it, so it fails at the arp level
<catern>
clever: oh nevermind I'm an idiot lol :)
<catern>
I tried so hard but
<catern>
I was on the wrong host and didn't realise it :)
<clever>
disasm: using ping like that forces it to not use a gateway
<clever>
disasm: to confirm, use a filter of 'arp or icmp' on tcpdump
<clever>
disasm: also, are you aware of how gateways work at the mac/ip layer?
<disasm>
clever: oh, you're right, that test isn't going to work at all...
<disasm>
clever: the basics of it, arp lookup for IP returns a mac, interface sends packets to that mac address, gateway then passes on packet to next gateway looked up in same way
<clever>
yeah, when using a gateway, you set the destination to the mac of the gateway, but leave the ip as-is
<clever>
so packets arrive at the gateway, with the gateway mac, but 8.8.8.8 as the ip
<disasm>
clever: yeah, that makes sense that ping ignores the gateway in that case... any tests I can do on the local box to make sure the gateway works before moving plugs around?
<clever>
the forwarding option in the kernel then tries to blindly obey the routing table
<clever>
and iptables mutates the from ip along the way
<clever>
manually plug 1 machine directly into the new router
<clever>
the wan side could be the original router, for double nat
<disasm>
right but that box has to have 802.1q interfaces enabled...
<dhess>
I wish someone would make an OpenWRT/LEDE-like distro using NixOS as a base.
<disasm>
dhess: nixos doesn't play very well with dynamic configurations :) You'd almost have to have a UI for writing configuration.nix and then having that run rebuild.
<disasm>
my goal is to get away from UI's anyways :) This router is deployed via nixops
<disasm>
I think after my wife gets off the phone (we have shotty cell service here so our phones are configured to default to wifi) I'm just going to go for broke and see what happens.
<dhess>
disasm: You don't have to do the config on the actual device running the LEDE-based NixOS. That could be a front-end thing that deploys to the target device, NixOps-style.
<dhess>
disasm: I don't think it needs to be a GUI, at least not at first.
Supersonic112 has quit [(Disconnected by services)]
<dhess>
disasm: are you using nftables on that router by any chance?
<disasm>
dhess: if you don't need a gui, just use straight nixops :) I'll be writing a blog post about my adventures here with my new apu.2c4 router on my company blog after I get everything stable!
ogle has joined #nixos
<disasm>
dhess: I don't think nftables plays well with networking.firewall at this point (don't see any integrations with a git grep at least)
Supersonic112_ has joined #nixos
<disasm>
so just plain iptables at this point
Supersonic112_ is now known as Supersonic112
drakonis has quit [(Ping timeout: 255 seconds)]
<dhess>
Yes, to use nftables you have to disable networking.firewall. At least last I looked at the manual that was true
* disasm
sad I'm going to have to get a new public IP :(
<ogle>
has anyone setup openldap? wondering if there is a guide somewhere, would like to get central authentication working on my boxes
<catern>
does anyone know of a supervisor with which to run nix-daemon, that supports monitoring to see if there is an active connection, and queues up restarts until the active connection closes?
<catern>
i'm setting up a large deployment of Nix and concerned about causing downtime for deploy scripts
<clever>
catern: under nixos, there is systemd socket activation, where systemd handles listening and creation of /nix/var/nix/daemon-socket/socket and it will launch nix-daemon upon the first connection
<clever>
catern: oh, and also, whenever you do connect to nix-daemon, it forks off a worker slave for that connection
<clever>
and the parent can die without impacting an in-progress build
<clever>
so you may not need to do anything
<catern>
clever: oh! huh! in that case then it should be fine then right
<catern>
clever: basically the master process only manages spinning off workers even without systemd
<clever>
yep
<clever>
as an example, check "ps -eH x" while a build is running
<catern>
neat neat neat
<catern>
nix sure is well designed
<clever>
the slaves also put the pid of their client into argv[1]
<clever>
so you can track down who is using that slave
<catern>
although i guess it's kind of annoying that the nix daemon will be down for an instant
<catern>
i guess if i was using systemd that would solve it
<clever>
yeah, no new connections can be formed during that time
<catern>
is it possible to set up that socket activation thing on non-NixOS?
<clever>
yeah, systemd would keep the socket open and queue them up, then relaunch the daemon
<clever>
should be trivial, just copy the nix-daemon.socket unit and nix-daemon.service
<clever>
just read the 2 files on nixos, and set them up such that they can run nix-daemon with the right env
<clever>
and i notice that the .service isnt directly tied to multi-user.target
<clever>
so the daemon doesnt run on boot, only te socket
<catern>
yeah, should be neat
<catern>
er, should work
<catern>
so I'll list that as a to-do, if and when I start running with systemd (I am currently not running nix-daemon as root but rather as a dedicated user)
<catern>
(I could investigate socket stuff with supervisord but, I kinda hate supervisord so I won't :) )
markus1199 has joined #nixos
drakonis has joined #nixos
<clever>
disasm: what have you done to me? lol, i was going to head to bed, now i need to render a graph of my entire network setup....
lambdamu_ has joined #nixos
markus1189 has quit [(Ping timeout: 240 seconds)]
lambdamu has quit [(Ping timeout: 248 seconds)]
<disasm>
clever: please share :)
<disasm>
clever: I'd love to see the source that generates the images as well :) I haven't played with dot yet. I've used dia in the past and latex as well for network diagrams.
nuncanada has quit [(Read error: Connection reset by peer)]
<clever>
and it updates the image you see in the gist! (just f5)
mbrgm has joined #nixos
<clever>
disasm: also, if you take a look at the graph, my nixos router is running a pppoe daemon on vlan 35, back-feeding into vlan 35 of the isp router
<clever>
thats because they havent documented how to configure things right
magnetophon has quit [(Remote host closed the connection)]
Dr8128 has quit [(Ping timeout: 240 seconds)]
JimBob has joined #nixos
dbmikus has quit [(Ping timeout: 240 seconds)]
sary has joined #nixos
simpson has joined #nixos
<simpson>
error: derivation ‘/nix/store/bzvk2l36w64byl6s8lvdffdpm3i387dz-openjdk-bootstrap.drv’ has incorrect output ‘/nix/store/52p11nv36hklpyw0fagff6qfrcw2jwcc-openjdk-bootstrap’, should be ‘/nix/store/f6iv68jw3cgbmgnyb70ks5pcx7wjs1kl-openjdk-bootstrap’
<simpson>
What's the right way to make Nix happy with me?
magnetophon has joined #nixos
justanotheruser has joined #nixos
<simpson>
Got it, just had to remove that derivation from the store.
justan0theruser has quit [(Ping timeout: 260 seconds)]
justanotheruser has quit [(Client Quit)]
justanotheruser has joined #nixos
magnetophon has quit [(Remote host closed the connection)]
Oida has quit [(Ping timeout: 255 seconds)]
magnetophon has joined #nixos
Oida has joined #nixos
astronavt has quit [(Remote host closed the connection)]
sigmundv_ has quit [(Ping timeout: 240 seconds)]
Arcaelyx has joined #nixos
Arcaelyx_ has quit [(Ping timeout: 255 seconds)]
CMCDragonkai1 has joined #nixos
CMCDragonkai1 has quit [(Ping timeout: 248 seconds)]
ylwghst has quit [(Ping timeout: 248 seconds)]
<fearlessKim[m]>
I have sandboxing enabled and ran nix-build --check on a package to rerun the build. The problem is that it doesn't rebuild (python) dependancies. Any way to propagate the --check ?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to release-17.09: https://git.io/vbRQQ
<srhb>
fearlessKim[m]: use nix-store -q --references /nix/store/your-derivation to get the immediate dependencies
grantwu has left #nixos ["WeeChat 2.0"]
<fearlessKim[m]>
thanks for the tip, it seems easier to force-delete the store entries
drakonis has quit [(Read error: Connection reset by peer)]
endformationage has quit [(Quit: WeeChat 1.9.1)]
DavidEGrayson has quit [(Remote host closed the connection)]
<fearlessKim[m]>
is there some doc on the new CI ? It fails with "Failed to enumerate outputs of master" which is too short for me
ma27 has quit [(Ping timeout: 258 seconds)]
stephaneyfx has quit [(Quit: Leaving)]
asuryawanshi has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
DavidEGrayson has joined #nixos
DavidEGr1yson has joined #nixos
DavidEGr1yson has quit [(Remote host closed the connection)]
DavidEGrayson has quit [(Remote host closed the connection)]
DavidEGrayson has joined #nixos
curry_nixos has joined #nixos
<curry_nixos>
anyone ever had a problem where everything in KDE is comically large after installing NVIDIA drivers?
<curry_nixos>
the screen resolution is correct, websites look good in firefox
Ivanych has quit [(Ping timeout: 268 seconds)]
<curry_nixos>
just menus, toolbars, buttons etc are all stupidly huge
Ivanych has joined #nixos
<fearlessKim[m]>
curry_nixos: could be a dpi problem ? maybe change dpi parameters via xrandr
MercurialAlchemi has joined #nixos
<curry_nixos>
fearlessKim: just xrandr --dpi with a bunch of settings and the screen flickered (so it seemed to be doing something) but no change
<fearlessKim[m]>
curry_nixos: no idea then sry
<curry_nixos>
fearlessKim: do you know how to make sense of the output of xrandr (with no options)?
<curry_nixos>
it says 160mm x 90mm, which if that's supposed to be the screen size then it's way too small
pie_ has quit [(Ping timeout: 255 seconds)]
<fearlessKim[m]>
these are pixels, but it 's still very low
curry_nixos has quit [(Quit: Konversation terminated!)]
Fare has joined #nixos
<fearlessKim[m]>
arf no I also see the mm size for the current screen, which is fine here
curry_nixos has joined #nixos
<curry_nixos>
it seems to be specifically only KDE-related stuff
<Fare>
how well does nixos work on ARM? nixpkgs? Would NixOS work on a Chromebook?
kreetx has joined #nixos
<curry_nixos>
i.e. Konsole has huge text, whereas xterm doesn't
<curry_nixos>
the parts of firefox that come from the OS GUI are huge, the web pages look fine
<fearlessKim[m]>
curry_nixos: silly questions but have you tried rebooting ?
<curry_nixos>
fearlessKim: always the right question to ask, but yeah i did just a couple minutes ago after playing with xrandr for a while. no change
<johnw>
does anyone know how to fix "ld: framework not found Cocoa" with Haskell projects on darwin? This didn't use to always happen, but now I can't build executables at all
<johnw>
or rather, it succeeds if Nix builds it, but fails if I nix-shell --command 'cabal configure' && cabal build
<Fare>
curry_nixos, is there a terminal emulator that doesn't suck?
<curry_nixos>
i could try installing one but it wouldn't fix the larger problem of everything being hilariously huge
curry_nixos has quit [(Quit: Konversation terminated!)]
curry_nixos has joined #nixos
<curry_nixos>
hey, problem solved! i had to set the dpi in the xserver settings in configuration.nix and now everything is wonderful
<curry_nixos>
fearlessKim: thanks for the dpi tip even though xrandr specifically didn't do anything
curry_nixos has quit [(Client Quit)]
pie_ has joined #nixos
<fearlessKim[m]>
When I run this in a shell it works fine, "nix-shell -p 'python3.withPackages(ps: [ ps.pygccxml ])' -I nixpkgs=/home/teto/nixpkgs3 " but as a shebang nothing. How can I diagnose shebang errors ?
<fearlessKim[m]>
curry_nixos cool
<etu>
Fare: I run stupidterm, it's easy to configure with a ini-file and is based on VTE :)
markus1189 has joined #nixos
stanibanani has joined #nixos
leat has quit [(Quit: WeeChat 1.9.1)]
markus1199 has quit [(Ping timeout: 240 seconds)]
<neonfuz>
trying to figure out how to get this usb to parallel cable to work
<neonfuz>
it's a prolific pl2302
<johnw>
nix-shell --command "cabal build" causes it to build, at least
<neonfuz>
can't figure out which module I need
pie__ has joined #nixos
asuryawanshi has quit [(Ping timeout: 248 seconds)]
pie_ has quit [(Read error: Connection reset by peer)]
cement has quit [(Ping timeout: 255 seconds)]
asuryawanshi has joined #nixos
kreetx has quit [(Ping timeout: 255 seconds)]
kiloreux has joined #nixos
<Lisanna>
sphalerite: I've seen tmux not update sometimes on the first attach, and so apps running inside it will be smaller than the pane
<Lisanna>
but doing a single size adjustment is always enough to make it update
zzamboni has joined #nixos
asuryawanshi has quit [(Ping timeout: 258 seconds)]
bert_ has joined #nixos
kiloreux has quit [(Ping timeout: 268 seconds)]
metaphysician has quit [(Ping timeout: 240 seconds)]
leat has joined #nixos
metaphysician has joined #nixos
asuryawanshi has joined #nixos
reinzelmann has joined #nixos
kiloreux has joined #nixos
Tucky has joined #nixos
<sphalerite>
Lisanna: yeah that's not what I'm getting
Itkovian has joined #nixos
ssmike has quit [(Ping timeout: 240 seconds)]
Itkovian has quit [(Ping timeout: 260 seconds)]
Itkovian has joined #nixos
tomster has joined #nixos
CMCDragonkai1 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vbRpQ
<NixOS_GitHub>
nixpkgs/master d1c9bee Franz Pletz: xmr-stak: fix evaluation with nix 1.11
NixOS_GitHub has left #nixos []
orivej has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to staging: https://git.io/vbRhe
<NixOS_GitHub>
nixpkgs/staging d945b3e Frederik Rietdijk: buildPythonPackage: remove bytecode from bin folder...
<NixOS_GitHub>
nixpkgs/staging 44f46a3 Frederik Rietdijk: Merge pull request #32542 from FRidh/bytecode...
NixOS_GitHub has left #nixos []
CMCDragonkai1 has quit [(Ping timeout: 260 seconds)]
Itkovian has quit [(Read error: Connection reset by peer)]
<manveru>
took some fiddling, but it'll take even more to actually build something with it :|
coot has joined #nixos
<hyper_ch>
viric: you're online?
<hyper_ch>
viric: I see you have packaged freeswitch for nixos....
Mateon3 has joined #nixos
<manveru>
freeswitch... good memories :D
Mateon1 has quit [(Ping timeout: 250 seconds)]
Mateon3 is now known as Mateon1
Itkovian has joined #nixos
<srhb>
Is there no path agnostic version of environment.etc ? I want to symlink in a file in /
<srhb>
I can do it in my service of course, but that seems ugly as well.
<tomster>
Mic92: so any chance we can get the omnibus issue fixed? like i said before i'm still new to both nixos *and* ruby but have to use chefdk here at work (where they only support ubuntu officially)... i'm eager to help just not sure where to start
<tomster>
i'm only using this because we're using chef at my new job...
<manveru>
yeah
<manveru>
well, i dodged that bullet, just thuoght i could maybe help :)
<tomster>
and all of my team is either on osx or ubuntu and i've been raving about nixos ever since i started and now i cannot run the test suite (which is written in chefspec)...
<manveru>
you tried running chefdk via their dockerfile?
<tomster>
the sad thing is that i got pretty much the entire rest of the stack running by now...
phdoerfler has joined #nixos
<manveru>
well, one more entry in my list of projects that bundix sucks at...
thc202 has joined #nixos
<tomster>
manveru: yes, this does seem more to be an issue with chefdk itself and less with nixos
<tomster>
my approach will be to might as well go ahead and learn myself some ruby (since i will need that to write chef and tests anyway)
ottidmes has joined #nixos
<tomster>
who knows, i might end up making a nixpkgs PR that fixes it :)
<srhb>
Is there a nice way of embedding a multi-line-string in a multi-line-string with indenting added to the inner one?
ob7 has joined #nixos
<ob7>
If a motherboard is running two identical GPUs, is it possible to load a different driver for each one?
zzamboni has quit [(Quit: Leaving.)]
<ob7>
I want to load a different driver for each one because nvidia driver works good in X but not so good in console. Since I mainly work in console, I need my console high resolution as well, and when using nvidia for X, console only capable of 1024 resolution. So if I can load nvidia on only one of the gpus, and let the console run on another, both my console and X session will look crisp. I would then run
<ob7>
another plug to the monitor and switch from DP to mDP to see X session and back to DP to work in console.
yenzenz has joined #nixos
jensens has quit [(Read error: Connection reset by peer)]
<ob7>
As for AMD gpu, I am able to work off a VEGA10 with both high resolution console and X session without needing two cards. The problem here is only the 4.15 kernel has monitor support at this time, and virtualbox is not building on 4.15 with amdgpu yet.
<sphalerite>
srhb: you mean so the *result* has indentation on the inner one?
<goibhniu>
ob7: I believe it's possible to use nvidia for one and intel for another using bumblebee (or similar) ... no idea about multiple nvidia cards or other variations
<ob7>
if anyone knows reply to ms5@ms5.io
ob7 has quit [(Quit: WeeChat 1.9.1)]
dywedir has quit [(Remote host closed the connection)]
bert_ has quit [(Ping timeout: 250 seconds)]
ilja_kuklic has joined #nixos
ottidmes has quit [(Quit: WeeChat 1.9.1)]
periklis has quit [(Ping timeout: 260 seconds)]
kiloreux has quit [(Ping timeout: 246 seconds)]
<hyper_ch>
doing now a test install with zfs.... using tank/boot for /boot and tank/encZFS/Nixos for encrypted root... let's see how that goes
b has quit [(Quit: Lost terminal)]
<sphalerite>
oooh, exciting!
bert_ has joined #nixos
<hyper_ch>
sphalerite: still I didn't just create the zpool running zpool create /dev/sda
<hyper_ch>
but I did first create a partition with fdisk... so that the first 2k sectors or something are empty for grub stuff
<sphalerite>
ah
<hyper_ch>
well, if that works I will try again using full disk :)
nschoe has joined #nixos
<hyper_ch>
weird error though: grub-install: error: unknown filesystem install-grub.pl: installation of grub on /dev/sda failed
pie__ has quit [(Ping timeout: 255 seconds)]
dywedir has joined #nixos
<srhb>
sphalerite: Yep!
<hyper_ch>
ok, I need boot.loader.grub.zfsSupport
<hyper_ch>
time to rebuild iso :)
<srhb>
sphalerite: But looks like it doesn't exist. I might create it.
<sphalerite>
Sounds like a plan!
<srhb>
What's the proper way of creating a symlink to a file outside of etc while a service is enabled?
<srhb>
I can do some fencing with preStart and postStop, but it's icky and could leak files.
FRidh has joined #nixos
<hyper_ch>
I'm still amazed that creating a custom bootable iso is so simple on nixos :)
<srhb>
Hrm. It's probably better to leak this file than to stick it in activation scripts.
pie_ has joined #nixos
<srhb>
Yuck.
<srhb>
Everything about packaging this module is horrible. :-P
ilja_kuklic has quit [(Ping timeout: 250 seconds)]
vandenoever has joined #nixos
vandenoever has quit [(Changing host)]
vandenoever has joined #nixos
asuryawanshi has quit [(Ping timeout: 246 seconds)]
<sphalerite>
srhb: what do you mean by it leaking files?
<srhb>
sphalerite: I mean when I disable the module, I want the symlink to go away. If it doesn't, I consider it leaky.
<sphalerite>
srhb: surely it'll still delete it when it stops the systemd service when switching to a config wit hteh module disabled?
<srhb>
(Contrast with the typical /var/lib/statefulModuleDir which should not be deleted)
<sphalerite>
the only way I see it not getting deleted is if you shut down uncleanly from a generation with the module enabled, and then boot into one with it disabled
<hyper_ch>
even when adding boot.loader.grub.zfsSupport to the iso and the actual configuration.nix for the new system, I still get the errors...
<hyper_ch>
too bad
<srhb>
sphalerite: Right, if postStop doesn't get executed, it's lost.
<sphalerite>
srhb: but under normal circumstanced postStop does get executed…
<sphalerite>
so it's not that leaky :p
himmAllRight has quit [(Ping timeout: 240 seconds)]
<sphalerite>
and even under abnormal circumstances it takes a fairly contrived situation to make it leak
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] phdoerfler opened pull request #32562: changed rm to rm -r in logcheck.nix (master...patch-1) https://git.io/vb0qe
NixOS_GitHub has left #nixos []
<srhb>
sphalerite: Fiiine. :-P
ilyaigpetrov has joined #nixos
<sphalerite>
srhb: or you could patch the software to have that file in /etc instead ;)
tmaekawa has joined #nixos
thblt has joined #nixos
tmaekawa has quit [(Client Quit)]
<phdoerfler>
Hi! I'm new to contributing to NixOS and my pull request failed checks (https://github.com/NixOS/nixpkgs/pull/32562): "Failed to enumerate outputs of master". Can I do anything about that?
<srhb>
sphalerite: That sounds scary, proprietary binaries and all.. I'm tempted though.
<sphalerite>
ooooh no wonder it's a pain x)
<phdoerfler>
this is I believe maybe the simplest possible pull request. To me this error looks unrelated to the "rm -r" I introduced
<sphalerite>
srhb: or you could stick it in a mount namespace and use a bind mount :D
<srhb>
phdoerfler: I don't think you need to worry about that. :)
<srhb>
sphalerite: That's... Also tempting.
<phdoerfler>
srhb: ok. Just making sure. What will happen next to this PR?
<srhb>
phdoerfler: Probably nothing since it's in 17.03 to be honest.
<phdoerfler>
srhb: I'm confused - I filed this PR against Master.
<sphalerite>
srhb: there's even systemd stuff for it, so that might even be the nicest option, see man systemd.exec, BindPaths=
<srhb>
phdoerfler: Oh sorry, I was looking at the issue.
<phdoerfler>
srhb: the issue very much still exists, someone else besides me ran into it recently, too and filed an issue.
<srhb>
phdoerfler: A member will eventually pick it up and decide whether to merge or not. If you want to speed it up, try figuring out which member is most likely to care about it (ie. git log the file and find a person that has contributed to the file and has merge rights)
<srhb>
phdoerfler: Tag them in the PR then :)
romildo has joined #nixos
<srhb>
phdoerfler: You could also test it.
<phdoerfler>
srhb: Thanks. I'll do that. However from my git blame I did it seemed like logcheck was last touched years ago.
<srhb>
phdoerfler: The checkboxes also tend to add additional confidence for potential mergers :)
<phdoerfler>
srhb: I suppose. But that's not likely to happen. I'm really new to NixOS and for me it was a choice between "submit a 0 effort PR" or "make your own derivation like the others did and not contribute"
sigmundv_ has joined #nixos
<phdoerfler>
srhb: this PR just changes a "rm" to a "rm -r" because it failed to delete a directory. That's all there is to it. So I thought I could get away without all these checkboxes
a6a3uh has joined #nixos
<phdoerfler>
srhb: you may call me lazy for doing so :)
pie_ has quit [(Ping timeout: 240 seconds)]
<sphalerite>
phdoerfler: they are by no means mandatory. But as srhb said they add confidence
charlycoste has joined #nixos
<sphalerite>
anyway, just give it a day or 3 and see :)
Mic92 has quit [(Quit: WeeChat 2.0)]
Mic92 has joined #nixos
<phdoerfler>
sphalerite: Thanks. I realize that and in fact I would try to tick some of them if I knew how to do any of them. But I have never made even an own derivation by myself or forked the nixpkgs repo and used that as a channel even, so there's that.
<sphalerite>
Yeah don't worry about it :)
<phdoerfler>
But I shall wait and see and be patient
<manveru>
don't think libXi is needed... will fix that
<makefu>
gchristensen: i really like how you (ab)use the github gist feature. lets see how long it takes until you get notified by github
<gchristensen>
:D
<gchristensen>
they haven't said anything to me yet, despite already creating 4,700 gists since like October
<manveru>
maybe you can clean them up after a few weeks?
<gchristensen>
that'd be polite
<manveru>
they probably don't like us very much already ;)
<gchristensen>
I always wondered what Travis thought about us
<gchristensen>
we don't use github like a CDN, like Homebrew and some windows package manager
b has quit [(Quit: Lost terminal)]
<manveru>
choco?
<manveru>
man... that thing was a mess :P
etu has quit [(Remote host closed the connection)]
etu has joined #nixos
<srhb>
If I have an option with type = listOf (submodule { ... }) and default = [ someValue ]; shouldn't the default merge strategy be to append to the list?
<gchristensen>
I think yes-except-excluding-the-default
<DIzFer[m]>
I'm pretty sure default is overriden: I recently stumbled upon this behaviour while making a module for default settings, with the networking.firewall.allowedTcpPorts list
<hyper_ch>
hmmmm, another microsoft tool got opensourced....
<etu>
hyper_ch: Weird world we live in...
<DIzFer[m]>
I used mkDefault, and ports were not opened
<hyper_ch>
not really sure what ProcDump is good for though
<gchristensen>
"Calamares is not a KDE project, or a KaOS project, or a Manjaro project, we are just a bunch of people dumping some code on GitHub." aren't we all ;)
griff_ has quit [(Quit: griff_)]
<disasm>
haha :)
<avn>
gchristensen: looks like project, where no one resposible for it ;)
grumble has joined #nixos
<manveru>
they're actually pretty helpful
<manveru>
but ofc our use-case is quite exotic
griff_ has joined #nixos
<sphalerite>
manveru: you were doing that web-based installer too, right? In some lisp?
<manveru>
yeah
zzamboni has quit [(Quit: Leaving.)]
<manveru>
started with guile+elm, migrated to go+elm, but eventually just gave up on that approach
<manveru>
don't have the time to do all this stuff myself
<manveru>
but i learned a ton through that, so pretty happy :)
<sphalerite>
That's good!
ma27 has quit [(Ping timeout: 276 seconds)]
<manveru>
the downside is that now i have to learn C++ and Python :P
<manveru>
which will take quite some time again...
<hyper_ch>
sphalerite: well, with apt, it can upgrade some packages and encounter problems with others and you're left in an unstable state... with nixos it first checks if config syntax still ok, then it builds stuff.. and only if everything was built find it makes a new generation entry.... so I think a stable channel could still be useful.
griff_ has quit [(Quit: griff_)]
<sphalerite>
hyper_ch: state can still be broken irreversibly (assuming no backups. Sure you have your zfs but it's a good assumption to make)
<hyper_ch>
what can break nixos?
proteusguy has joined #nixos
<clever>
if /nix/var/nix/db/db.sqlite is deleted, you basically have to start over
<sphalerite>
Changes in how stateful data is managed from one version to the next
<hyper_ch>
I was told nixos is can't be broken :)
<hyper_ch>
(well, it's hard to break it)
<clever>
hyper_ch: there was also a bug a year ago in the nixpkgs channel (not the nixos channel), that corrupted the grub.conf file
<clever>
the nixos channel didnt update, because the testing caught it and blocked the channel
<clever>
then all the users on the wrong channel came out of the woodwork :P
<hyper_ch>
I'm sure you just wanted to see which users use the wrong channel and give them a wakeup call ;)
<sphalerite>
Yes it's hard to break, but it does require some common sense and some sensible decisions on the part of the maintainers - like not providing a stable channel
<gchristensen>
!m sphalerite
<[0__0]>
You're doing good work, sphalerite!
<sphalerite>
^^
szicari has joined #nixos
crucialrhyme has joined #nixos
<hyper_ch>
sphalerite's mojo just got a huge boost :)
<crucialrhyme>
i'm just getting started with nixos on my home box (so i actually own it and it will probably never be a multi-user system)
<crucialrhyme>
should i mainly be doing everything in configuration.nix, or mainly doing everything via nix-env, or something else?? or??
<hyper_ch>
sphalerite: why isn't everyone using zfs?
<hyper_ch>
crucialrhyme: I prefer everything in the configuration.nix
<hyper_ch>
also single-user system
<manveru>
crucialrhyme: i usually try quick stuff with nix-env, and clean it up after a while, putting stuff i actually use into configuration.nix
<hyper_ch>
with shotgun security system
<hyper_ch>
I like the configuration.nix route because I can track it with git
<clever>
manveru: i tend to test things with nix-shell
<clever>
manveru: even after removing it from nix-env, the rollbacks keep the package around for ages
<hyper_ch>
clever: I thought you had your own datacenter for testing?
<crucialrhyme>
cool, thanks! configuration.nix it is
<crucialrhyme>
at least for the important stuff
<manveru>
hm true
<clever>
hyper_ch: more for seeing if a program is even worth using and keeping long-term
<crucialrhyme>
i already have a nix-shell going with CUDA and pytorch and some other ML stuff and i was shocked at how easy it is
<hyper_ch>
clever: tried to get /boot into a zfs pool but it didn't work out :(
cmiles74 has joined #nixos
<crucialrhyme>
i guess it's also like, normally i'd install as much as possible to my home dir to avoid "polluting" the global directories with garbage that might cause compatibility programs later
<crucialrhyme>
but that's not a problem with nixos!
<clever>
crucialrhyme: it is with QT based apps
<sphalerite>
hyper_ch: licensing issues which make it harder to distributee, greater complexity, applications where it's not suitable (e.g. low-end devices that don't have enough RAM)
<crucialrhyme>
clever: what do you mean by that?
<clever>
crucialrhyme: QT expects to find some of its libs in a place like ~/.nix-profile/lib/ and if you install apps from 2 different qt versions, they break eachother
<hyper_ch>
sphalerite: :) but it's so nice
<hyper_ch>
anyway, I'm glad I migrated my windows vm from vbox to kvm :)
<clever>
crucialrhyme: the only solution ive found is to upgrade both apps, so they are on the same QT version
<crucialrhyme>
would it work to not install either globally and run each in a separate nix-shell?
<clever>
crucialrhyme: hmmm, not sure if it would find the libs at all then
<clever>
crucialrhyme: installing them globally in configuration.nix would also fix it, because then they always come from the same nixpkgs, and the same QT version
<clever>
so no conflict can exist
<crucialrhyme>
oh i see
<clever>
its more an issue about using qt-app-1 from 5 months ago, with qt-app-2 from yesterday
<hyper_ch>
yey to hash-based paths :)
<hyper_ch>
how is actually the store-path being generated? it's some kind of hash but not sure what it's based on
<sphalerite>
hyper_ch: it is indeed! Maybe one day we'll have nixos on kfreebsd which eliminates the licencing issues ;)
<sphalerite>
Maybe.
<crucialrhyme>
i'm pretty pleased with nixos so far and it's something i wish more scientists were aware of (not just FP nerds and sysadmins). there's a big push for reproducible research, but the way people do it basically involves putting everything in docker containers
<crucialrhyme>
it seems to solve the same problem only way better
<sphalerite>
hyper_ch: the derivation and the output name, or the contents hash and the name in the case of fixed-output derivations
zzamboni has joined #nixos
<hyper_ch>
sphalerite: :)
<toogley>
can i use a branch inside fetchFromGithub?
<manveru>
yeah... the benefit docker has is that you don't actually have to be rigorous :P
<toogley>
haven't found sth in the nixos manual
<toogley>
eeh nixpkgs
<hyper_ch>
toogley: you can provide that just commit
<hyper_ch>
besides that I don't know
<toogley>
hyper_ch: aah, how do i do that?
<clever>
crucialrhyme: i have also been using nix for a lot of data processing tasks, i split each stage of processing into its own nix derivation, one to parse the data into a db, one to sort, filter, and print latex, another to turn the latex into a pdf
<clever>
crucialrhyme: then i can work on each stage seperately, and if i find a bug in an early stage, nix will automaticaly re-run all stages in the right order
<crucialrhyme>
manveru: as soon as you try to build a nontrivial docker image you end up in dependency hell and sorting it out is at least as painful as being rigorous would be
<clever>
crucialrhyme: nixpkgs also has docker utils, one minute
<manveru>
crucialrhyme: which is why i build docker images via nix ;D
<manveru>
clever: yeah, that's basically what i do, based on LnL's nix-docker stuff
<manveru>
but it's not trivial to understand for newbies
griff_ has joined #nixos
<clever>
crucialrhyme: lines 153-158 are an extra bit to make it hydra friendly, you now have a simple download image link on https://hydra.iohk.io/build/56778
<crucialrhyme>
if you all happen to know of similar tools for working with singularity i would be even more pleased
<clever>
so you can pipe that directly into `docker load` and not even need nix
<clever>
'The Singularity software can import your Docker images without having Docker installed or being a superuser.'
<gchristensen>
use nix to create a docker image, then use singularity's docker-to-singularity converter?
<clever>
crucialrhyme: it might be able to just convert the docker images
<crucialrhyme>
yeah it def can
<crucialrhyme>
in terms of like the system. but there's config stuff around it too
<LnL>
manveru: about that, do you guys also use it to run tests etc?
<manveru>
LnL: yes
<toogley>
hyper_ch: ah, thanks
<LnL>
manveru: how do you handle caching
<manveru>
well, it's one persistent docker running
<manveru>
from there i build the images, and start test-scenarios using docker-compose
<LnL>
oh, not a fresh container for every build
<manveru>
no... that was way too slow
<clever>
yeah, i would make a persistant container with sshd and nix-daemon, that just acts as a dumb build slave
<LnL>
yeah...
<manveru>
exactly :)
<clever>
you could even launch it with `docker run --rm` so the store is lost at shutdown
<clever>
the darwin host will copy things into the container again at the next build
<clever>
and the docker image itself can be build with nix
<LnL>
well, the usecase I'm looking for is using cached nix dependencies in a container that might run impure stuff
<clever>
ah, i was thinking more about a generic linux build slave
<clever>
rather then impure linux actions
roberth has joined #nixos
ylwghst has quit [(Remote host closed the connection)]
<crucialrhyme>
another dumb question is, sometimes sites (esp cross-platform graphical programs like IDEs or whatnot) will offer like a .deb or .rpm for download in the browser, outside of the system package repositories. what would the nixos equivalent of this be, if there is one?
ylwghst has joined #nixos
<joepie91>
crucialrhyme: there was once a one-click-install-y thing I think, but I don't think it's used anymore
<magnetophon>
I'm packaging a go program and one of the goDeps wants pkgconfig. Should I make a separate pkg for that and include it?
<hyper_ch>
someone in here has written that really great wiki entry :)
<joepie91>
crucialrhyme: the problem is that such an 'out of band' package fits more into an imperative package management model than a declarative one
<joepie91>
because imperative packages are, when it comes down to it, just tarballs with some metadata that get dumped on disk and maybe deps installed
<clever>
you could always download the default.nix for the package and nix-env -f default.nix -i
<joepie91>
crucialrhyme: probably the closest thing in NixOS that doesn't break the declarative model, are overlays
<goibhniu>
crucialrhyme: you can install a package from the binary cache if you know the store path, which you can find on hydra, the build farm (this is similar to the one-click-install, except without the button)
<crucialrhyme>
it seems like either overlays or that could work depending on the use case
<sphalerite>
What I usually do nowadays for stuff that I don't plan to integrate into nixpkgs is write an overlay and a nix expression that imports nixpkgs with that overlay for convenience so I don't depend on the (stateful) installation of an overlay
<crucialrhyme>
for like, RStudio or something, an overlay would probably make more sense (except that one is already in nixpkgs, which is excellent)
<manveru>
anyone happen to use emacs (spacemacs) and python?
babyflakes has quit [(Quit: Connection closed for inactivity)]
bfrog has joined #nixos
<magnetophon>
manveru: I sometimes try to. Have you got the side-window working? All I get is errors.
<manveru>
heh, just trying for the first time
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 2 new commits to master: https://git.io/vb0Si
<tomster>
it hast the concept of a omnibus_root directory where all the packages dump their stuff into
zzamboni has quit [(Client Quit)]
<gchristensen>
can the installer be run during nix-build time?
<tomster>
gchristensen: i guess so
<gchristensen>
then set the omnibus_root to $out
<tomster>
gchristensen: my approach would indeed be to use nix to build a package that satisfies omnibus' expectation
iyzsong has quit [(Ping timeout: 240 seconds)]
<tomster>
gchristensen: the problem is that the directory inside the nix store ($out) is missing things
<tomster>
mainly the embedded/bin directory
<tomster>
the 'embedded' directory seems to be the dumping ground where omnibus dumps everything
<tomster>
it does not exist anywhere on my machine, particularly not inside the chefdk store entry
<tomster>
it seems to expect that the ruby its being run with is the embedded ruby that chefdk brings with itself and expects its own root directory to be ../../.. of the location of the ruby it's running
Sonarpulse has quit [(Ping timeout: 255 seconds)]
<gchristensen>
jinkies. that is nasty :/
<tomster>
so i have two problems: first the relative location and be the (lacking) contents of that location
<tomster>
gchristensen: yes and i thought python packaging was messy...
zzamboni has joined #nixos
<tomster>
the thing is, their approach works for them and they claim to support all three platforms (linux, osx and windows -> case closed)
<gchristensen>
well, they _do_
dbmikus has joined #nixos
<tomster>
i'm not sure how to continue... either give up on nixos for work and install ubuntu (which would feel dirty by now) or somehow make omnibus nixos compatible
<gchristensen>
just not all linuxes :|
<gchristensen>
why do you need chefdk specifically? I wonder if you could create your own "chefdk" using the same Gemfile they do and nix's ruby infra
<gchristensen>
chefdk exists because package management on other linuxes sucks
<tomster>
gchristensen: well i need to use chef exec etc...
<tomster>
gchristensen: you mean chefdk is just a bundle of ruby gems?
<gchristensen>
yeah
<tomster>
with its own installer?
<gchristensen>
yeah
<tomster>
but how would i track releases then?
<manveru>
pretty sure you can run chefdk if you patch out their omnibus stuff...
<tomster>
are we talking about a 'source' install of chefdk>
<manveru>
tomster: you could do it with a custom gem config via bundlerEnv
<tomster>
gchristensen: but isn't that exactly what the nix pkg for chefdk is doing anyway?
<manveru>
i can maybe give it a quick try...
<gchristensen>
tomster: no
<manveru>
oh there's one :)
<tomster>
because the software itself is installed just fine
ylwghst has quit [(Remote host closed the connection)]
<tomster>
it's just that at runtime it makes the assumption that it has been omnibusinstalled
<gchristensen>
chefdk is built with `omnibus` which is a nix lookalike which rebuilds all software in a subdir isolated just for their omnibus builds
<tomster>
gchristensen: as i have learned earlier today...
cmiles74 has quit [(Ping timeout: 240 seconds)]
<tomster>
and omnibus is a) completely unaware of nixos and b) conceptually incompatible with nixos
<gchristensen>
so patching the omnibus outputs to look like nix paths is going to be very difficult
marl has joined #nixos
<tomster>
gchristensen: exactly
<gchristensen>
well conceptually very SIMILAR to Nix, which is part of why it is so hard to fix :)
<tomster>
well *you* are the expert on that question :)
<marl>
How do I login to postgres? I just enabled it and it says that I am supposed to have access according to the manual. I tried user postgres and root on 17.09.
<gchristensen>
I am? :$
<marl>
It tells me that role "root" or "postgres" don't exist in those cases.
freusque has quit [(Quit: WeeChat 1.9.1)]
<manveru>
tomster: i dunno man, but in your case i'd just run chefdk in docker...
<tomster>
marl: FWIW i'm on 17.09 too and "psql postgres" works for me OOTB
<gchristensen>
tomster: I'd listen to manveru, as the knower of ruby things in nixpkgs.
<marl>
tomster: that can't be possible.
<tomster>
manveru: well the problem is that we're using to run kitchen whcih in turn uses virtualbox to fire up VMs
<tomster>
you can't do that inside docker...
<marl>
tomster: for me it says that user <my user> does not exist.
<tomster>
marl: for me, too
<manveru>
oh fun
<joepie91>
marl: what do your postgresql logs say?
<tomster>
marl: you need to create a postgres user yourself
<joepie91>
they usually give more detail about login failures
<tomster>
marl: does "psql postgres" work for you or not? it's not quite clear from what you posted
cement has joined #nixos
<tomster>
manveru: it's really frustrating i've been stuck at work with this for the second day now
<tomster>
and my project is lingering...
griff_ has joined #nixos
<gchristensen>
tomster: :( that is very frustrating
<marl>
tomster: psql: FATAL: role "myuser" does not exist
<marl>
tomster: it does not.
simpson has left #nixos ["WeeChat 1.0.1"]
<tomster>
marl: that is expected
<tomster>
marl: i'm not asking about <myuser> i'm asking you about "psql postgres"
mkoenig_ has joined #nixos
<marl>
tomster: trust me, I am answering your question. I am running psql postgres
<tomster>
marl: ok, but then what about <myuser>?
mkoenig has quit [(Ping timeout: 248 seconds)]
<marl>
tomster: I am executing this as user myuser.
<domenkozar>
marl: by default you can login only through root
<domenkozar>
you'll have to change pg_ident for user logins
<domenkozar>
iirc
<domenkozar>
tomster: hey!
<tomster>
domenkozar: are you sure?
<tomster>
domenkozar: hey!
<marl>
domenkozar: same result: psql: FATAL: role "root" does not exist
<manveru>
tomster: well, i can give it a try, but don't have a ton of time since i have to fly tomorrow
<tomster>
i just installed postgres two weeks ago and the postgres user worked OOT for me
<tomster>
manveru: what would you try in this case?
<domenkozar>
marl: try psql -U postgres
<marl>
domenkozar: psql: FATAL: Peer authentication failed for user "postgres"
<toogley>
can i force a rebuild via nix-env ? when i do nix-env -e pkg, then it will not be rebuild
<domenkozar>
tomster: nix-shell needs a file or -p
<tomster>
domenkozar: ah, right
babyflakes has joined #nixos
<tomster>
i'm doing that inside my project
<tomster>
wait i will post my default.nix
<marl>
domenkozar: the manual is like 15 lines of text/virtually non-existent.
<tomster>
domenkozar: -p chefdk should do the trick
<tomster>
marl: did you try my snippet? works for me...
<marl>
tomster: your snippet is the same as what 17.09 does, looking at the source code from what I have seen, but perhaps I looked at nixpkgs master and not 17.09.
<marl>
tomster: I can try it, but I don't think I should require that.
<tomster>
marl: i have since switched to unstable but it worked with 17.09 OOTB
<tomster>
marl: sorry i cannot be more helpful... it simply works for me :-/
<marl>
The issue is that the manual never has instructions specifying a step like "you can now login".
<tomster>
that's what the default.nix says how it created the package
<marl>
tomster: psql -U postgres postgres works with your configuration.
* gchristensen
covers his face in shame
<marl>
The current 17.09 postgresql module is broken.
griff_ has joined #nixos
<tomster>
very helpful, that's how i upgraded it from 2.3.4 to 2.4.17 :)
<tomster>
i did a PR and it has already been merged
<tomster>
and the week before that you upgraded it from 1.x to 2.3.4 for me after i asked about it here :)
<tomster>
you probably don't remember because you do this stuff probably every day for a ton of people...
<tomster>
marl: ah, good to know
<marl>
Is there some way to do declarative users in configuration.nix for postgres too?
<tomster>
marl: then i probably did do some other stuff afterwards
<marl>
I would like to have a plain user to also have some access.
<tomster>
marl: FTR i use plain old .sql files for that
<domenkozar>
seems like the problem is that GEM_HOME is not set
<tomster>
domenkozar: yes, and that a gem home does not exist on nixos
periklis has quit [(Ping timeout: 260 seconds)]
<tomster>
at least not a single one
<tomster>
AFAIK
<tomster>
at least on macos that /opt/chefdk/embedded/bin directory contains a whole lot more stuff than the bin directory inside the chefdk nix store
<tomster>
domenkozar: but maybe setting GEM_HOME to the nix store's one would be enough
ylwghst has joined #nixos
thblt has quit [(Ping timeout: 260 seconds)]
griff_ has quit [(Ping timeout: 260 seconds)]
<manveru>
tomster: the problem is that the chefdk has a helpers.rb with usr_bin_prefix hardcoded
<manveru>
which must be /usr/bin on linux
<domenkozar>
wooot, ghcjs compiles!
<toogley>
do you have tips for debugging: "error: derivation ‘/nix/store/2slc85kf64dc3qds3lfa61ch7msfnr2n-neomutt-20171208.drv’ may not be deterministic: output ‘/nix/store/519cp7axvh8s6d8bxwbif7ki09a422kb-neomutt-20171208’ differs" ? i have set nix.useSandbox = true;
<manveru>
so i'd patch helpers.rb, the `usr_bin_path` method should instead just be: `def usr_bin_path(command); `which #{command}`; end`
<manveru>
which may be hacky, but should at least work :)
<gchristensen>
toogley: /opt/chefdk/embedded/bin contains _everything_ it needs to run, which obviously nix won't do
<domenkozar>
manveru: but nothing calls usr_bin_path
Neo-- has quit [(Ping timeout: 240 seconds)]
leat has joined #nixos
<marl>
toogley: something might be inserting a timestamp or something like that.
<manveru>
the chef executable calls verify!
<manveru>
which does use usr_bin_path
<domenkozar>
it does? I don't see it in traceback
<toogley>
gchristensen: can you explain what that means regarding my issue? i can't follow your thoughts
<gchristensen>
sorry toogley
<toogley>
marl: ah, thanks
<gchristensen>
tomster: /opt/chefdk/embedded/bin contains _everything_ it needs to run, which obviously nix won't do
<toogley>
gchristensen: aaah :D
<manveru>
oh, nevermind...
<manveru>
it has `return true unless omnibus_install?`
<manveru>
which in our case is false
<gchristensen>
toogley: are you specifically trying to make neomutt a binary-deterministic build?
<tomster>
gchristensen: yes, i have told you already as much earlier
<tomster>
gchristensen: i understand and appreciate the problem and have done my homework so to speak i just don't know how to approach this problem conceptually let alone specifically
<gchristensen>
sounds good, I hope you can solve it
<tomster>
i know that it's not just a matter of bending or setting the path
<manveru>
tomster: so the issue is `File.expand_path(File.join(Gem.ruby, "..", "..", ".."))`?
<disasm>
Well, got a good start on my new nixos router last night :) Only thing not working is ipv6 routing. I'm even getting delegation from ISP, and radvd is broadcasting everything fine: https://github.com/disassembler/network Big thanks to gchristensen for his inspired network nixops repo as well as clever for helping me debug some non-issues last night :)
<tomster>
manveru: precisely
zzamboni has quit [(Quit: Leaving.)]
<manveru>
and you tried fixing that line?
<tomster>
manveru: a) that it's the wrong (relative) path
<gchristensen>
hot dog, disasm!
<toogley>
gchristensen: no, not at the moment. but i know already what the reason is. our docs are not reproducible, they contain timestamps
<tomster>
and b) that under nix no such difrectory exists
<gchristensen>
toogley: ok, then you don't need to fret about that problem
<disasm>
I already factory reset my lede asus rt-n16 so there's no going back now (not that I'd want to)
<marl>
disasm: on what platform will this run?
<tomster>
it's the directory where it expects all of its binaries to reside
<domenkozar>
tomster: I don't see what installs /opt/chefdk/embedded/bin
<toogley>
gchristensen: thanks
<marl>
disasm: MIPS32 by any chance?
<disasm>
marl: see the github repo, I link the device on the readme
<tomster>
and under nix these are spread out among multiple nix store entries
<disasm>
marl: it's x86_64
<gchristensen>
toogley: it is just saying it rebuilt the exact same neovim again and it came out differently
<tomster>
domenkozar: nothing in the nix pkg
<tomster>
domenkozar: the nix pkg doesn't use omnibus
<manveru>
tomster: i'd probably replace omnibus_env then
<tomster>
but chef expects to have been installed via omnibus
<disasm>
marl: I think your on your own for mips right now. If you want to improve it though, I'm sure the community would love it! https://nixos.wiki/wiki/NixOS_on_MIPS
<toogley>
gchristensen: neomutt, you mean?^^
<manveru>
tomster: basically tell it to stop babysitting you
<toogley>
gchristensen: but thanks for the explanation^^
<tomster>
while i'm giddy with excitement and have the same warm fuzzy feeling about the nix community that domenkozar has been raving about two years ago when he first told me about nixos... what do i need to do now to get it running for me? :-P
<tomster>
manveru: i have my default.nix that i enter with a nix-shell that points to a local clone of nixpkgs
<phdoerfler>
domenkozar: this looks useful, thanks!
<disasm>
marl: my guess is there'd be a lot of the bootstrapping of tools and some cross compilation fun. If you're interested, might want to talk to copumpkin that did the darwin stuff.
<manveru>
tomster: the difference then is that i did that in a checkout of chef-dk repo
<tomster>
manveru: also i would like to contribute to fixing the chefdk pkg instead of just getting it work for me
<manveru>
aye
<manveru>
well, lemme quickly patch the package then
erictapen has quit [(Ping timeout: 260 seconds)]
<tomster>
manveru: awesome!
<phdoerfler>
manveru: Just discovered nixos-option, actually :) However it seems to dump the content of my config in a way that bash tries to parse it. It tries to call it as commands
<tomster>
manveru: and i will learn from it and try to upgrade it for the next release myself
playX has quit [(Ping timeout: 260 seconds)]
<sphalerite>
marl: I'd say it depends on a lot of things — especially what you need working. I suspect (but I'm not sure) that I personally might be able to get the essentials running in two weeks of work or so. But if we're talking about a router with 64MB of flash memory for the OS that won't fly for example, it would require a USB device or something to run the root filesystem off or something
<sphalerite>
I got nixos running on my chromebook over a couple months of (very part-time) tinkering, but that's ARM which is more widely supported as well
<marl>
sphalerite: I think the target for something like that should compete with OpenWrt (or its name du jour).
<marl>
sphalerite: and that's 16MB of flash.
<marl>
sphalerite: two weeks is impressive.
<sphalerite>
marl: yeah… *that's* not going to happen with nixos any time soon
<marl>
sphalerite: what's the largest issue there?
<sphalerite>
marl: it's just a guesstimate, I'm not very experienced with embedded systems and it depends on how friendly the hardware is to work with as well
<marl>
sphalerite: I meant space wise.
<sphalerite>
marl: nixos depends pretty strongly on glibc and systemd for now
dbmikus has joined #nixos
<marl>
sphalerite: pluggable libc and pluggable init system.
<marl>
sphalerite: yeah, those are pretty big.,
Tucky has quit [(Quit: WeeChat 2.0)]
<sphalerite>
yeah, there are people who want to do that but it won't be working anytime soon AFAIK
dbmikus has quit [(Client Quit)]
seanparsons has quit [(Ping timeout: 276 seconds)]
dbmikus has joined #nixos
<sphalerite>
OTT, does the church of suckless nixos have any updates anywhere?
<tomster>
manveru: so would i simply need to replace the chefdk pkg's gemset with the contents of yours?
<manveru>
tomster: no
<manveru>
it's actually missing some stuff
<tomster>
manveru: i'm a bit lost on how to get it working for me
<manveru>
give me a bit of time man :P
<tomster>
the pkg?
<tomster>
manveru: oh, you're on it?
<manveru>
working as fast as i can :)
<tomster>
thanks man
<tomster>
manveru: don't :)
seanparsons has joined #nixos
<tomster>
that's too fast :)
<manveru>
lol
aminechikhaoui has quit [(Ping timeout: 248 seconds)]
<codedmart>
How do you update npm? If I try the usual `npm i -g npm` I get a permission error.
<playX>
try with sudo?
<codedmart>
Is that the recommended way?
<sphalerite>
no and it won't help
<playX>
i don't know i'm using pip, but that only way i know
<sphalerite>
playX: please don't give advice like that
<sphalerite>
playX: sudo is only the appropriate solution for some permission errors. In many others, it's likely to screw permissions up on whatever you're installing. Especially on nixos, sudo is often not what you want.
<avn>
playX: does you definelly need playonlinux? Just wine not works for you?
<playX>
avn: from playonlinux more easily install dependencies for game, and fallout 2 from wine don
<playX>
launching
<sphalerite>
playX: Usually the only cases where sudo *is* the right thing to use is applying changes system-wide (e.g. nixos-rebuild on nixos, apt on debian-likes, systemctl for service management etc) and for modifying system-wide configuration (stuff in /etc, usually nothing other than /etc/nixos on nixos)
<avn>
playX: easiest way is look what fallout2 need, and install it via winetricks
Neo-- has joined #nixos
<playX>
sphalerite: yes, i know sudo in nixos it's not the best way, but how to update package if it's only older version in channel
<playX>
avn: choosed system version wine, it's finded 0_o
<avn>
build package from master, or just overwrite src in package
<manveru>
dunno... my patches entry doesn't do anything...
<avn>
playX: it should work as long, as you not try to build patched wine with PoL
tusj has joined #nixos
<sphalerite>
playX: you can use the latest version from nixpkgs-unstable, and if that doesn't have it submit a PR updating it to nixpkgs
<avn>
Folks, anyone familiar with scsi controllers for modern (pci-e) machines?
<hodapp>
SCSI? what year is it?
* hodapp
rubs eyes
<playX>
sphalerite: i'm already using 18.03 version:)
aminechikhaoui has quit [(Ping timeout: 260 seconds)]
aminechikhaoui has joined #nixos
<sphalerite>
playX: so you update it in nixpkgs yourself and submit a PR :)
<avn>
hodapp: want to attach old scanner to nixos powered machine ;)
<hodapp>
why a SCSI one >_<
<gchristensen>
avn: wow :D
<playX>
sphalerite: if i'm add older channel with kde 4 can i use kde4 with 18.03 version, or it's impossible?
<avn>
hodapp: because it costs me 40$ (where usb one costs $400) -- is a film/slide scanner
<avn>
(well, with delivery and card/cable I hope to fit to $150)
dbmikus has quit [(Ping timeout: 250 seconds)]
dbmikus_ has joined #nixos
<hodapp>
what kind of resolution & dynamic range for film/slides?
<avn>
But yes.. I know more about scsi stuff, that about pci-e, and some stuff like 1x/4x scary me
<hodapp>
does SANE claim to support it?
<avn>
yes
JosW has joined #nixos
<hodapp>
and you've only PCI Express, no PCI?
<thoughtpolice>
Dezgeg: Hey, sorry for the ping -- I remember you recently got QEMU patches upstream to run generic armv7l NixOS images -- are there any notes on doing this by chance? I know of your wiki page but was looking into build these images myself.
<avn>
hodapp: I have one PCI, but my next machine wouldn't have it. So I thinking about 1x pci-e atm. adaptec 29320 looks funny and enough cheap.
<hodapp>
looks funny?
<Dezgeg>
thoughtpolice: if you have irc logs for this channel, grep for qemu-system-arm or '-machine virt' or something
<gchristensen>
avn: I think if you need 1x then anything more than 1x will be sufficient
rotaerk has quit [(Ping timeout: 246 seconds)]
<avn>
hodapp: Just kidding. It looks like "should works".
<hodapp>
"2017-11-28 19:10:15 Dezgeg you need a monster command like qemu-system-arm -machine virt,highmem=off -cpu cortex-a15 -bios ~/opt/u-boot/build-qemu_arm/u-boot.bin -s -nographic -drive if=none,file=$HOME/armtest.qcow2,id=root -device ich9-ahci,id=ahci -device ide-drive,drive=root,bus=ahci.0 -netdev user,id=net0 -device virtio-net-device,netdev=net0" - this?
<Dezgeg>
yes
<avn>
gchristensen: think so. Will also check, if sane need update, and do it if need ;)
<gchristensen>
:)
<Dezgeg>
eep, that should not be using virtio-net-device though
<avn>
will back to it, when my scanner arrive from US ;)
<thoughtpolice>
hodapp: Dezgeg - Thanks! Was just about to grep myself, I'll look through the relevant convo.
<thoughtpolice>
Would this be worth adding to the NixOS on ARM wiki page, or is it all a bit too in-flux still? ISTR the needed patches being upstream in U-Boot/Qemu already?
<Dezgeg>
doc/README.qemu-arm in u-boot repo has the official docs
<Dezgeg>
yes it's all in upstream u-boot, save for a one-liner to bump up the kernel command line length limit for nixos
<hodapp>
I flip-flop between whether I want to try to run NixOS on my Pi, or stick with Raspbian and get the better support
<Dezgeg>
so yeah, the wiki could be updated
<playX>
guys, i'm added nixos 15.09 repository to use KDE4 how i can add option to configuration.nix to use kde4?
<thoughtpolice>
Dezgeg: cool, thanks! I should be off to the races, then, I think. I'll look into adding wiki notes when I have a chance.
<gchristensen>
sorry, we don't support 15.09 and aren't inclined to help you do that, playX.
sigmundv_ has quit [(Ping timeout: 250 seconds)]
griff_ has quit [(Ping timeout: 240 seconds)]
danl1240 has joined #nixos
periklis has quit [(Remote host closed the connection)]
erictapen has joined #nixos
charlycoste has quit [(Quit: Leaving.)]
griff_ has joined #nixos
rotaerk has joined #nixos
<marl>
Is it possible to build one NixOS system on a list of build servers?
sbjorn has quit [(Ping timeout: 255 seconds)]
coot has quit [(Read error: No route to host)]
<marl>
playX: man configuration.nix RET / kde4 RET
coot_ has joined #nixos
<avn>
playX: also `git grep kde4` in checked-out sources can give you similiar result
<marl>
playX: running KDE4 over KDE5 won't get you anything, though.
ma27 has joined #nixos
<manveru>
tomster: sorry, i was fighting my own stupidity :P
<tomster>
manveru: i added your fork of nixpkgs as a remote to mine and switched to your chefdk branch
<tomster>
then i re-entered my shell using the same local path. it downloaded and installed quite a bit (as expected) and then simply entered `chef exec` as a simple test and then got the same error
<tomster>
manveru: should i be doing something differently? were you able to get `chef exec` working?
chpatrick has joined #nixos
<Synthetica>
What is the recommended way to make multiple versions of a package available (both 32/64/rpi and stable/beta)
<Synthetica>
If the build procedure is (basically) the same
<chpatrick>
do you know how I can get a nix shell with a non-nixpkgs package installed?
<chpatrick>
specifically I have a haskell executable package defined with callCabal2nix, and I'd like to use this executable without installing it
<chpatrick>
similar to -p but for a custom package
<Synthetica>
Can't you use `-I`?
<tomster>
manveru: oops, my bad, i must have not properly exited the nix-shell previously (noticable by still having chefdk 2.4.17 in the paths)
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
<tomster>
manveru: basically chef exec gives /nix/store/bqs4bfdrwgdc67iwvg5fc7w5l9sm83h4-chefdk-2.4.21/lib/chef-dk/command/exec.rb:30:in `exec': wrong number of arguments (given 0, expected 1+) (ArgumentError)
datakurre has quit [(Ping timeout: 255 seconds)]
<tomster>
and even just running `chef --version` errors out (with /nix/store/wfymk1bq7m5zdwk5js8viand68s48630-chefdk-2.4.21/lib/ruby/gems/2.4.0/gems/mixlib-shellout-2.3.2/lib/mixlib/shellout/unix.rb:340:in `exec': No such file or directory )
datakurre has joined #nixos
<tomster>
manveru: were you able to run `chef --version`? how did you test? i'm leaving work now, i will keep an eye on your repo - i'm more than happy to test or help in any other way. thanks for your effort so far!
aminechikhaoui has joined #nixos
[0x4A6F] has joined #nixos
Dezgeg has quit [(Ping timeout: 264 seconds)]
tomster has quit [(Ping timeout: 258 seconds)]
aminechikhaoui has quit [(Ping timeout: 268 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 8 new commits to master: https://git.io/vbEtJ
<NixOS_GitHub>
nixpkgs/master cb6725e Peter Simons: LTS Haskell 9.17
<NixOS_GitHub>
nixpkgs/master 3ddc73a Michael Alan Dorman: hackage2nix: keep old haskell-src-exts_1_19_1 around...
<NixOS_GitHub>
nixpkgs/master 75cc0ff Peter Simons: hackage2nix: add old hpack version to make stack-1.6.1 happy
NixOS_GitHub has left #nixos []
Dezgeg has joined #nixos
yenzenz has quit [(Ping timeout: 248 seconds)]
seequ has quit [(Ping timeout: 248 seconds)]
seequ has joined #nixos
playX has quit [(Quit: Page closed)]
goibhniu has quit [(Ping timeout: 268 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] dtzWill opened pull request #1723: Fix forward-decl of Bindings as "class", match definition. (master...fix/bindings-struct-class) https://git.io/vbEtN
<srhb>
Judson: That path is missing a leading slash. However, you're probably getting an unexpected version of man, and that's why the db is weird.
<srhb>
Judson: But you can try adding in the slash first.
<Judson>
Yeah...
<Judson>
Leading slash was the problem.
<Judson>
Sorry, went into human terminal mode
<srhb>
S'ok, it's difficult to proofread oneself :-P
<manveru>
tomster: you have to give an argument to exec
kreisys has joined #nixos
<manveru>
tomster: like chef exec ls
nevermind has joined #nixos
nevermind is now known as Guest48699
Guest48699 is now known as nevermind
ChongLi has joined #nixos
CMCDragonkai1 has quit [(Ping timeout: 258 seconds)]
Dezgeg has quit [(Ping timeout: 240 seconds)]
dywedir has joined #nixos
cement has quit [(Ping timeout: 255 seconds)]
seequ has quit [(Ping timeout: 255 seconds)]
seequ has joined #nixos
<Judson>
I might've finally puzzled out tmux + direnv forever...
<clever>
?
<Judson>
Putting `set-environment -gu DIRENV_...` for DIFF WATCHES LAYOUT and DIR in to .tmux.conf
cement has joined #nixos
ixxie has joined #nixos
Dezgeg has joined #nixos
<Judson>
Otherwise tmux copies the DIRENV stuff into its global env and copies those into every new window.
<Judson>
Relevant mostly because of nix-shell
<clever>
ah
<clever>
i havent used direnv, and i prefer screen over tmux
<Judson>
So your "?" was mostly "why are you mentioning that here?"
<Judson>
Why do you prefer screen?
Neo-- has quit [(Remote host closed the connection)]
<Judson>
(I need to enumerate and record the reasons I switched)
<clever>
i learned screen first, and now all the keybinds are muscle memory
<clever>
and tmux seems to lack some basic features like forcibly resizing the window when others are viewing
Neo-- has joined #nixos
<Judson>
Fair. The majority of my conf is "make it work like screen" :)
<clever>
or swapping the windows within splits
<srhb>
clever: It has the former at least.
<srhb>
Not sure about the latter.
<clever>
for screen, the splits are at a seperate level, and each buffer can switch to any tab
<srhb>
(But yeah, why switch)
<clever>
for tmux, the splits are inside the buffer
<clever>
and i'm constantly re-organizing what things go in which split
<srhb>
Yes, I've always wondered over that abstraction as well.
<clever>
sometimes i want to see 2 things at once, sometimes i want the editor to take 100%, without closing the 2nd thing
<Judson>
tbh, I rarely use screen/tmux splits
<clever>
i dont use a tiling window manager, so all tiling happens within screen
<srhb>
I also tend to end up with multiple terminals sharing the same tmux session, which invariably results in four terminals on the same workspace showing the exact same thing at times.
<srhb>
It's quite hypnotic and, well, useless.
* srhb
needs a better terminal workflow
<clever>
screen also has a - buffer, that just shows nothing
<clever>
so you can de-focus all buffers, without actually disconnecting
<Judson>
srhb, that happens to me sometimes but usually when I'm doing something weird. How does it happen all the time?
<Judson>
clever, what's the use case for that?
<srhb>
Judson: My default usage is... What's the terminology again... One shared super session with all terminals connected.
<srhb>
Judson: So all terminals can pick one window from the same pool of windows
<clever>
Judson: i connect to screen from many terminals and many computers
<srhb>
Judson: I painstakingly arrange which windows are active in which terminal.
<clever>
but it only auto-resizes if there is only 1 terminal focused
<srhb>
Judson: This works fine until I start shutting down windows, and then multiple terminals are on the same window.
<srhb>
Like I said, it's a poor workflow.
<Judson>
Huh.
<Judson>
I use tmux mostly for development, and use one session per workspace.
<clever>
Judson: also, there is a massive cpu hit if the buffer doesnt match the screen size
<clever>
Judson: if the sizes match up, screen/tmux can just relay the bytes directly to the terminal, and its fast
coot_ has quit [(Quit: coot_)]
<Judson>
I see, so you want e.g. xterm to resize to the screen buffer.
<clever>
but if the buffer is smaller, it has to track the column, and insert control characters to sek a lot
<Judson>
Where tmux does the opposite.
<clever>
and if the buffer is bigger, it has to redraw the entire screen as it pans
<clever>
tmux forcibly fits the smallest viewer, and doesnt have any bypass
<clever>
screen has ^aF to force it to fit the viewer using that hotkey
<clever>
another major difference/anoyance
<clever>
i have 5 terminals open with screen, each showing a different buffer
<clever>
tmux cant mux!!
<clever>
all viewers must always see the same thing
ma27 has quit [(Ping timeout: 255 seconds)]
<sphalerite>
clever: yes it can! It shows the output of multiple terminals on one!
ma27 has joined #nixos
<srhb>
It can also show different windows in different terminals of course. :P
Itkovian has joined #nixos
<sphalerite>
srhb: nope, only if they're in different sessions
Jackneilll has joined #nixos
<sphalerite>
that's what clever's complaining about :p
<srhb>
sphalerite: What's wrong with different sessions though?
<clever>
srhb: i want to dynamically change what i see in each terminal, depending on the situation
<clever>
without having to disconnect and connect to a new session
<sphalerite>
srhb: more of a pain to switch between than windows
<srhb>
huh?
<sphalerite>
and you don't have a convenient list of them along the top
<sphalerite>
(or bottom as the case may be)
<sphalerite>
clever: C-b s
<srhb>
This is what new-session -t sess does
<sphalerite>
you don't have to disconnect to switch sessions
<srhb>
So two terminals with new-session -t0 will see the exact same possible windows but don't have to view the same ones.
<clever>
sphalerite: ^a5 is simpler
<sphalerite>
srhb: huh, really?
<srhb>
sphalerite: Yes, that's my workflow.
<clever>
sphalerite: and what about moving the bottom split into its own terminal?
<clever>
srhb: oops, ^^^
<Judson>
terminal or window?
<sphalerite>
srhb: neat! Although it still forces the size for all the windows :(
yenzenz has joined #nixos
<srhb>
clever: No, as I said earlier that's one thing I'm flabbergasted about as well in the tmux abstraction.
<Judson>
Because you can for sure move panes to their own window etc
<srhb>
Splits are just.. Weird.
Jackneill has quit [(Ping timeout: 240 seconds)]
<srhb>
If you can move it to its own window, you can move it to its own terminal
<srhb>
(with the new-session groups I described)
<Judson>
Still I recall being frustrated with tmux's model a couple of times
<clever>
there are only 2 things tmux has that i want in screen
<clever>
verticle splits, and being able to sometimes control what a remote viewer sees
<Judson>
Copy mode can't be "left" on an unfocused window?
<sphalerite>
clever: what are they? Because i'm just wondering why I don't use screen
<Judson>
i.e. scroll back into a backtrace, then try to bring up windowlist...
<srhb>
Judson: What?
<sphalerite>
Judson: yes it can
<clever>
sphalerite: let me screenshot an example
<cement>
does the install script require root?
<sphalerite>
Judson: or do you mean exited by left?
<srhb>
cement: The nix install script? Yes, it will prompt you.
<NixOS_GitHub>
[nixpkgs] lheckemann opened pull request #32574: Correct comment in radicale.nix (master...radicale-comment-fix) https://git.io/vbEcX
NixOS_GitHub has left #nixos []
KenVogel has joined #nixos
drakonis has quit [(Read error: Connection reset by peer)]
<cement>
he figured I might need the other tools (and things like llvm) because I'm needing to do some dev on this machine
<sphalerite>
btw how long does the nixos-small channel take to build and test from scratch on a recent, fairly powerful machine?
<sphalerite>
hehe
<srhb>
sphalerite: Server or consumer?
<sphalerite>
"nope, nix will do! :D"
KenVogel has quit [(Client Quit)]
<cement>
yeah, I haven't actually seen nix's list of deps, so I just let him go ahead
<sphalerite>
srhb: consumer. Specifically it's a Dell XPS 15, 9560
<sphalerite>
cement: you shouldn't need anything unusual to install nix using the install script once you've got your /nix, be it with superuser help or with nix-user-chroot
<sphalerite>
just bash and curl really
<sphalerite>
and coreutils
<sphalerite>
since nix will live in the store it'll have all its deps in the store too
pie_ has joined #nixos
evangeline has quit [(Ping timeout: 255 seconds)]
<clever>
bbl
ilyaigpetrov has quit [(Quit: Connection closed for inactivity)]
<srhb>
sphalerite: iirc on small machines the main issue is having to run all the vms sequentially. I don't recall how many are in the small channel
<srhb>
sphalerite: But they suck up a lot of memory.
<sphalerite>
I don't need precise estimates, but would like to know whether to expect 1h, 5h, or 20h
<sphalerite>
ah right
<sphalerite>
I've got 16GB in here
sbjorn has joined #nixos
<sphalerite>
ooh, I could make use of remote builds too though, got access to a server that also has 16GB
<srhb>
It's just nixos, not nixpkgs, right?
nixer has joined #nixos
<srhb>
A few hours probably then.
<sphalerite>
cool, thanks
<nixer>
I forget, how do I call nix-shell against a standard callPackage default.nix file in nixpkgs?
<srhb>
If it's in nixpkgs and has an attribute the easiest way is nix-shell '<nixpkgs>' -A attribute
<srhb>
A top level attribute, that is
<srhb>
if not, something like nix-shell -E 'with import <nixpkgs> {}; pkgs.callPackage ./path/to/default.nix {}'
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bjornfor opened pull request #32575: projectlibre: init at 1.7.0 (master...projectlibre) https://git.io/vbEWz
NixOS_GitHub has left #nixos []
<srhb>
Hm, I still have no idea how to do -E kind things in nix 1.12
<srhb>
Nor how to step through a build with the usual nix-shell '<nixpkgs>' -A attr
i-am-the-slime has joined #nixos
evangeline has joined #nixos
<Judson>
(Is the talk about nix 1.12 up somewhere?)
<efx[m]>
what is the recommended way to copy a package and its dependencies to a file but not using nix-copy-closure? I'm getting an error when loading a list of imported derivations that were exported via nix-store --export
<srhb>
efx[m]: Did you do the recursive references thingamajick when exporting?
<gchristensen>
use nix-store --export with $(nix-store -qR /nix/store/...-the-package-you-want-to-send)
<srhb>
^ that
<gchristensen>
use `nix-store --export $(nix-store -qR /nix/store/...-the-package-you-want-to-send)`
<efx[m]>
txs, I'll try with the nix-store -qR
<efx[m]>
I was using nix-store --export $(nix-store --query --requisites $(type -p $base) "$other")
<Judson>
Thanks both.
<efx[m]>
from the docs, but that could be out of date?
Oida has quit [(Ping timeout: 240 seconds)]
oida has joined #nixos
<efx[m]>
well, I am actually passing 2 paths to the nix-store --export so that may not work
<srhb>
efx[m]: It should, as long as their requisites are included, I think.
<efx[m]>
as an aside, one of the packages is a compiled library and does not have a predictable binary path
<efx[m]>
so passed the raw path to nix-store ~/.nix-profile/lib/mod_security2.so
<srhb>
efx[m]: There's nothing wrong with the import here, is there?
<srhb>
efx[m]: It's just saying it doesn't know which of those to stick in your profile.
<efx[m]>
hmm, not sure. I grabbed that list from what nix-store --import returned, so maybe I shouldn't paste all of it?
<srhb>
What are you actually trying to accomplish? I'm very confused with the httpd.closure part
<srhb>
I think we need to step back a bit :-)
Dezgeg has quit [(Ping timeout: 246 seconds)]
<efx[m]>
ha, thanks for the help. No worries. I would like to install httpd and modsecurity in a nix-env, then export to a single closure I can install to another machine that has nix.
<efx[m]>
basically a manual nix-copy-closure
<srhb>
Sounds like a job for buildEnv
Synthetica has quit [(Quit: Connection closed for inactivity)]
<efx[m]>
huh, so whip up some nix expressions with that function?
<srhb>
I think so, yes
marl has quit [(Quit: Page closed)]
<efx[m]>
thanks for the tip. at a high level, I want the assurance all subsequent nodes run the binary compiled on the first node
<srhb>
efx[m]: At least I can do something like nix-build -E 'with import <nixpkgs> {}; pkgs.buildEnv { name = "modsecurityttpd"; paths = with pkgs; [ modsecurity_standalone apacheHttpd ]; }'
<srhb>
But I'd just transfer the individual ones if they aren't connected.
<cement>
so... the nix install script automatically sets the default/only channel to nixos-unstable? why is that?
<efx[m]>
awesome thanks for the help srhb
<srhb>
cement: Most of the extra stable tests are mainly relevant for NixOS. They're mostly tests of services and so on
ylwghst has quit [(Remote host closed the connection)]
<srhb>
cement: So I guess it's a nice balance between new and tested :)
ylwghst has joined #nixos
<srhb>
cement: Scratch everything I said there that's plain wrong.
GiGa has joined #nixos
GiGa has quit [(Changing host)]
GiGa has joined #nixos
<srhb>
The tests are the same, the branches are different.
* srhb
is tired clearly
<GiGa>
Evening all (substitute timezone relevant greeting)
<srhb>
GiGa: Evening.
dywedir has quit [(Remote host closed the connection)]
<GiGa>
Hey srhb, how's things?
<srhb>
GiGa: Good thanks, and you?
kreisys has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
leat has quit [(Ping timeout: 248 seconds)]
mizu_no__ has joined #nixos
Dezgeg has joined #nixos
<GiGa>
srhb, very well thanks. Been doing some interesting forensics at work.
<srhb>
:) Always fun.
endformationage has joined #nixos
<Lisanna>
nix-prefetch-git doesn't seem to work with git SHA1s of branches? It works if the SHA1 is of a commit that's in the master branch.
<sphalerite>
oooh, forensics?
<GiGa>
There wasn't too much challenge sadly
<sphalerite>
Lisanna: it should. Can you provide an example of where it's failing for you?
ma27 has joined #nixos
<Lisanna>
hmm, it *might* be because it's pulling from a checked-out git repo which might not have the branch information downloaded? I don't know how git works, but I know that "git log origin/<branch>" works
<Lisanna>
pulled from a checked-out git repo = "--url file:///path/to/repo"
<Lisanna>
But, origin has: master (da50ac644) and branchfoo (da7070fbb), I git clone the repo, get the sha1s of master and branchfoo using git log, and then pass the sha1s to nix-prefetch-git
<Lisanna>
nix-prefetch-git accepts the master sha1, but rejects the branchfoo sha1
<Lisanna>
let me try checking-out the branch first, so that it creates a local version of the branch
<Lisanna>
git's CLI is so weird...
<clever>
Lisanna: the root problem, is that the git protocol doesnt allow you to fetch a given sha1, you must fetch a branch
<clever>
Lisanna: its a security thing, to protect deleted branches i think
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
<clever>
there are comments in the fetchgit source explaining it
<Lisanna>
Okay, so checking out the branch first should fix this
<Lisanna>
since I'll have a local copy of the sha1
<sphalerite>
ugh, when do we get a matrix client that supports encrypted chats and has reasonable performance
<Lisanna>
yep, that worked
<jeaye>
Is there a nix approach to running services within nix-shell? For example, I want a nix-shell for development which is also running postgres as per some configs. I can see this working with containers or docker, but I wonder if there's a lighter way.
Guanin has joined #nixos
ericsagnes has quit [(Ping timeout: 240 seconds)]
<srhb>
It doesn't get much lighter than NixOS containers, I think.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fare opened pull request #32576: directvnc: 0.7.5-test-051207 -> 0.7.7-test-20150416 (master...directvnc) https://git.io/vbERP
NixOS_GitHub has left #nixos []
<srhb>
The problem is that you need the module system, or you'll have to do all the bootstrapping yourself.
<srhb>
Which means NixOS.
GiGa has left #nixos ["Leaving"]
GiGa has joined #nixos
<jeaye>
I'm on NixOS, which is no problem. Can containers be specified outside of configuration.nix, on a per-user level?
GiGa has left #nixos ["Leaving"]
GiGa has joined #nixos
<srhb>
They can be created imperatively, but I think you do need root.
<jeaye>
I'd like it to be declarative, so it seems like configuration.nix is my only choice.
<srhb>
jeaye: You can make it quite modular with imports.
<jeaye>
It is already. :)
<srhb>
I meant containers specifically :-)
<jeaye>
Yep, I do that as well.
<jeaye>
Sharing modules between the host and the containers.
* srhb
nods
<jeaye>
It's kind of a pain, since containers don't inherit all of the good security settings set in the host.
<Guanin>
Hi, is there any way to cleanup the system, so that it is in a state similar to directly after nixos-install? (I'm playing around in a VM and when I change a secret for gitlab or do some other changes that create a database or similar unpure data, stuff breaks)
<jeaye>
So that sort of modularity is required, I think, to sanely configure containers.
<srhb>
jeaye: Yeah, I feel you.
<srhb>
Guanin: It's specific to the module in question unfortunately. For gitlab, you can nuke /var/gitlab and /var/lib/postgresql or whatever it's called (bearing in mind you'll lose all other databases as well)
<clever>
jeaye: there is also the service runner script, one minute
<jeaye>
I wonder if there could be some system for more easily specifying the inheritance of properties in containers.
<Guanin>
srhb, nuking those dirs does not cause a recreation of the required data until I change something related in the configuration.nix again
<srhb>
Guanin: The creation is tied to the systemd services in both cases.
<srhb>
Guanin: So stopping them, nuking them and restarting them should suffice.
<clever>
jeaye: create a configuration.nix that fully describes the service you want to enable, then build this, nix-build -I nixos-config=./configuration.nix '<nixpkgs/nixos>' -A config.systemd.services.postgresql.runner
<srhb>
clever: TIL!
<clever>
jeaye: that will generate a perl script, that will handle pre/post start, start, and pre/post stop
<Guanin>
Thank you srhb, this might help a lot :)
KenCoder has joined #nixos
<jeaye>
clever: How would that work with nix-shell?
<jeaye>
It would require manual `systemctl start ...` right?
<LnL>
I didn't think it works for postgres
<clever>
jeaye: the runner script has zero involvement in systemd
duncan^ has quit [(Changing host)]
duncan^ has joined #nixos
<clever>
jeaye: so you could refer to that perl script in the shellHook for the nix-shell env
<LnL>
it doesn't understand runAsRoot
ssmike has quit [(Ping timeout: 268 seconds)]
duncan^ has quit [(Quit: WeeChat 1.9.1)]
globin has quit [(Ping timeout: 240 seconds)]
<jeaye>
clever: I see, thank you. That's a neat approach.
<jeaye>
Under different circumstances, I'd call it clever.
<clever>
:D
duncan^ has joined #nixos
ericsagnes has joined #nixos
leat has joined #nixos
Neo-- has quit [(Remote host closed the connection)]
Neo-- has joined #nixos
ssmike has joined #nixos
ssmike1 has joined #nixos
ssmike has quit [(Read error: Connection reset by peer)]
ssmike1 is now known as ssmike
leat has quit [(Ping timeout: 248 seconds)]
ssmike has quit [(Ping timeout: 240 seconds)]
zzamboni has joined #nixos
<srhb>
Hmm. What might be causing the bad mtimes in tests/boot.nix that fail hard after 574526d510 ?
<srhb>
I'm unsure how to debug.
civodul has joined #nixos
ssmike has joined #nixos
benasse has joined #nixos
<gchristensen>
srhb: I think it is because of a problem in the hydra builderns
<srhb>
gchristensen: That's the motivation for the commit, right? The error is reproducible locally.
<gchristensen>
right
<gchristensen>
hrm
coot_ has joined #nixos
<srhb>
For instance it builds /nix/store/vdz6jzhm6fjx3djgcxcxp4g1i1zcfp1p-etc-host.conf with an mtime of Nov 12 (with current nixos-unstable)
kreisys has joined #nixos
kus_ubuntui686 has joined #nixos
kus_ubuntui686 has quit [(Max SendQ exceeded)]
<LnL>
isn't that the iso build issue?
<srhb>
Oh there's an issue for it?
simendsjo has quit [(Ping timeout: 248 seconds)]
<srhb>
I must have not searched for the right words
mizu_no__ has quit [(Quit: Computer has gone to sleep.)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<srhb>
I'm only surprised because that test has failed since the commit on dec 3 and I'd seen no mention of it anywhere.
<srhb>
And reverting it seems to "fix" the test though that's probably not desirable :-P
<srhb>
I wonder how long those mtimes have been screwy.
zzamboni has quit [(Quit: Leaving.)]
<sphalerite>
surely those shouldn't be able to exist in the first place, nix should have set their mtime to 0?
<srhb>
1, right? But yes.
whatnoise has joined #nixos
<srhb>
Probably something's been broken for a long time in the test infrastructure. I don't see it for similar files on my host system.
<srhb>
LnL: That's the issue that lead to the fix that breaks the (broken) test.
<srhb>
I think.
asuryawanshi has quit [(Ping timeout: 240 seconds)]
<LnL>
oh, for some reason I thought you where talking about december last year :p
<srhb>
Oh sorry. :)
<LnL>
dunny why
<srhb>
It's really peculiar. The wrong mtimes all appear to be around Nov 12 12:02-12:03 CET
<srhb>
I feel like I'm being fed broken store paths from the cache or something
ixxie has quit [(Quit: Lost terminal)]
<srhb>
Is that even possible?
<LnL>
hmm, possibly
<sphalerite>
Surely not? I don't think nars have a field for time
<clever>
oh, thats why my browser has been acting wonky
<clever>
hydra started to build ghc, and i ran out of hdd space entirely
<clever>
so hard that i cant even garbage collect
<srhb>
sphalerite: Hm.. I wonder where that time is from then.
<LnL>
are you using nix 1.12?
<srhb>
Nope
<LnL>
it reserves some space in the statedir now so it can still gc even when the volume is at 100%
<toogley>
is it sane to use a path like /run/current-system/sw/bin/sendmail in a nixos pkg?
<clever>
LnL: zfs is immutable, you need free space even to delete a file
<srhb>
Uuuh
<clever>
LnL: because it has to create a new version of the directory without that file
<LnL>
ah
<srhb>
Okay, if I delete this store path and then nix-store --repair-path it, it returns with the same weird Nov 12 timestamp
* srhb
boggles
<sphalerite>
toogley: no, typically you'd want to either have a hard dependency on something that provides sendmail or invoke it from the PATH so that nix-shell and user profiles can affect it too
<clever>
LnL: the only way out is to manualy destroy a snapshot, then GC can work
<LnL>
isn't that only the case when you have snapshots?
<clever>
LnL: i have snapshots on everyting
<clever>
except /nix itself
zzamboni has quit [(Quit: Leaving.)]
<LnL>
I have the reverse, everything except for the store since I can rebuild that
<clever>
and also for error recovery after an improper shutdown, it never does in-place edits on things like a directory listing
<clever>
/nix is the only thing not covered, because the snapshots make GC useless, and it can always be rebuilt
<LnL>
I did add /nix/var recently, after I corrupted my db :p
<clever>
maybe next time, i'll leave /nix/var on the / dataset, and use /nix/store/ for the non-snapshot one
<toogley>
/usr/sbin/sendmail with some valid nixos sendmail path, if some sendmail binary is installed. so how can i test if some package has installed sendmail, if i can't use /run/current-system/sw/bin/sendmail ?
<clever>
LnL: also, silly java is throwing a FileNotFoundException, with a message of no space left on device, lol
<LnL>
heh
<srhb>
I'm afraid some file system weirdness is going on that I don't understand. If I delete the file and restore it with nix-store --repair-path, it returns with the same weird mtime. if I then set its mtime to 1, delete it and restore, its mtime stays at 1.
<srhb>
Such confusion.
kreetx has joined #nixos
<clever>
srhb: what exactly is the FS setup?
<srhb>
clever: /nix and /nix/store and both separate datasets in my only pool
<srhb>
rpool/store /nix/store zfs defaults 0 0 according to fstab
<clever>
but when nix passes it to systemd, it borks hard
<clever>
as in, systemd murdered every process and dropped me into an emergency shell :P
<LnL>
I mean quoted for systemd fileSystems."\"foo bar\"" = {};
aarvar has joined #nixos
<clever>
i think it would be more user-friendly if nix just unconditionaly quoted everything
<LnL>
yeah, definitively :)
<clever>
but i havent risked my system stablity to test variations
thblt has joined #nixos
<sphalerite>
toogley: I'm a bit confused, by the looks of it it's only used in init.h… and there it's concatenated with some flags and then the pointer converted to an integer!?
<whatnoise>
Sorry to intrude, and my apologies for my newbiness.. I'm having an issue where systemd hangs at boot. (new install) looking into the journal the only issue I see that failed was 'backlight:acpi_video0' no such device. after view fixes I tried appending boot.kernelParams = [ "acpi_backlight=vendor" ]; into my configuration.nix but still the same issue.. any ideas as to what else I can do?
<sphalerite>
toogley: but by the looks of it it's a setting that can be set at runtime as well, and apparnetly it's passed through a shell or at least a word splitter at some point along the line. So at least one of "sendmail" and "/usr/bin/env sendmail" should work.
<sphalerite>
toogley: I'd consider the former neater if it works.
<sphalerite>
whatnoise: how does it hang?
<whatnoise>
sphalerite: it just freezes. even after the 90sec kill nothing else happens. i cant even open a shell
<toogley>
sphalerite: so you think my initial approach to use substituteInPlace is fine? just use /usr/bin/env sendmail
<sphalerite>
whatnoise: does it respond to ctrl-alt-del, power button, magic sysrq?
dbmikus__ has joined #nixos
<sphalerite>
toogley: yeah that should be fine I think. Unless autosetup has a proper way to set the variables manually in which case that would be preferabale
<toogley>
sphalerite: ok, thanks
dbmikus_ has quit [(Ping timeout: 240 seconds)]
<sphalerite>
toogley: perhaps configureFlags = ["SENDMAIL=sendmail"]; would be enough
<whatnoise>
sphalerite: to ctrl+alt+del and power button only
<sphalerite>
(or similarly passing SENDMAIL=sendmail to ./cofigure in some other way)
ylwghst has quit [(Quit: Lost terminal)]
<akfp``>
how can I have nix-build keep all build directories?
<akfp``>
or at least the last one?
<sphalerite>
whatnoise: that suggests that systemd itself is fine, but that some unit isn't completing its start.
<srhb>
Resetting the store mtimes by hand fixed the issues. Go figure.
<sphalerite>
whatnoise: grep the journal for "Start" and see if you can find any "Starting …" with no mathcing "Started …"
<sphalerite>
akfp``: I usually use --keep-failed and deliberately cause the derivation I'm interested in to fail
<sphalerite>
akfp``: e.g. by adding false at the end of its installPhase
stanibanani has quit [(Ping timeout: 248 seconds)]
Lisanna has quit [(Ping timeout: 260 seconds)]
<Guanin>
I got gitlab running (still in a VM) and it complains that I should "update asap". I am on 10.0.2 and use the 17.09-small channel. Is this correct?
<whatnoise>
sphalerite: the onlyone without a matching started is Starting Load/Save Screen Blacklight Brightness of backlight:acpi_video0
dj_goku_ has joined #nixos
dj_goku_ has quit [(Changing host)]
dj_goku_ has joined #nixos
<whatnoise>
sphalerite: backlight:radeon_b10 starts thereafter successfully though. then nothing else in the log
szicari_ has joined #nixos
<sphalerite>
whatnoise: weird. Maybe set `systemd.services."systemd-backlight@backlight:acpi_video0".enable = false;`
hiratara has quit [(Ping timeout: 258 seconds)]
<sphalerite>
although if it actually failed, as opposed to hanging, idk if that would make a difference.
<sphalerite>
Worth a shot nonetheless I guess.
orivej_ has joined #nixos
cement_ has joined #nixos
ajp_ has joined #nixos
ciil_ has joined #nixos
fgaz_ has joined #nixos
<whatnoise>
sphalerite: hmm ill try. would i add that to the nixos/configuration.nix ?
AtnNn_ has joined #nixos
oleks_ has joined #nixos
<whatnoise>
well looking at my fail log i can see it did fail
kosmikus1 has joined #nixos
<whatnoise>
perhaps i can mask i mask it?
<sphalerite>
yep
nahamu_ has joined #nixos
<sphalerite>
Also since systemd itself isn't hanging you may also want to try booting into single-user mode, by adding systemd.unit=rescue.target on the kernel command line (by pressing E on the GRUB menu or systemd), and see if that'll get you a shell
nmikhailov_ has joined #nixos
hiratara has joined #nixos
GiGa has quit [(Quit: Leaving)]
joncfoo_ has joined #nixos
hyper__ch has joined #nixos
swflint_away has joined #nixos
goodwill_ has joined #nixos
ldleworker has joined #nixos
tylerjl has joined #nixos
marcinkuzminski_ has joined #nixos
andymandias_ has joined #nixos
Fuuzetsu_ has joined #nixos
nmikhailov_ is now known as nmikhailov
hyper__ch is now known as hyper_ch
goodwill_ is now known as goodwill
Fuuzetsu_ is now known as Fuuzetsu
swflint_away is now known as swflint
szicari_ is now known as szicari
bert_ has quit [(Ping timeout: 260 seconds)]
infinisil has joined #nixos
nevermind has quit [(Read error: Connection reset by peer)]
metaphysician has quit [(Ping timeout: 240 seconds)]
<sphalerite>
whatnoise: yes, that's what the .enable = false; thing I wrote earlier does
hamishmack has quit [(Quit: hamishmack)]
vandenoever has joined #nixos
vandenoever has quit [(Changing host)]
vandenoever has joined #nixos
metaphysician has joined #nixos
pikajude has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
pikajude has joined #nixos
<whatnoise>
sphalerite: neither worked. pressing e failed as well hmm
roberth has joined #nixos
simukis has quit [(Ping timeout: 255 seconds)]
imagio has joined #nixos
<catern>
i am a gen-i-us
bgamari has joined #nixos
dan_b has joined #nixos
i-am-the-slime has quit [(Ping timeout: 276 seconds)]
<sphalerite>
whatnoise: pressing e on the boot menu didn't do anything?
<sphalerite>
and did you rerun nixos-install after making the change to configuration.nix?
fgaz_ is now known as fgaz
kreetx has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm pushed 1 new commit to master: https://git.io/vbEiS
<whatnoise>
sphalerite: pressing e didnt do anything no. which is very odd
[0x4A6F] has quit [(Ping timeout: 276 seconds)]
<sphalerite>
whatnoise: are you using systemd-boot or grub?
digitus has joined #nixos
<whatnoise>
ill rebuild nixos once i can connect through ethernet. as my wifi drivers wont work. (another problem for another day lol)
<whatnoise>
systemd-boot
<sphalerite>
did you set boot.loader.systemd-boot.editor to false?
<sphalerite>
catern: what did you do?
<whatnoise>
its set to true
ldleworker is now known as ldlework
ldlework has quit [(Changing host)]
ldlework has joined #nixos
dan_b has quit [(Ping timeout: 276 seconds)]
nisstyre has quit [(Changing host)]
nisstyre has joined #nixos
<sphalerite>
yeah that seems sensible. Do any keys work in systemd-boot?
<whatnoise>
besides the enter key or ctrl+alt+del no.
<whatnoise>
i might just do a clean install.. i feel like its odd behavior. i have nothing to lose as its a new install anyway
<sphalerite>
huh. idk, I'm at a loss
<whatnoise>
its alright. thanks anyway :)
Neo-- has quit [(Remote host closed the connection)]
Neo-- has joined #nixos
<catern>
sphalerite: I wrote a script which walks a derivation and impurely builds any fixed-output paths found in its inputs
<sphalerite>
catern: neat! with nix-build --hash?
<sphalerite>
or rather: that sounds like it would be good with nix-build --hash
<catern>
it's a generic replacement for nix-prefetch-url, -git, -svn, -ftp, etc... it's also very useful for me, whose nix-daemon can not access the outside world :)
bert_ has joined #nixos
<catern>
sphalerite: yes
<catern>
though I'm not actually using the output hash
<catern>
I just like the side-effect of getting these things into the store :)
<sphalerite>
well that's what nix-build --hash does too :p
<catern>
i am using nix-store --hash actually
<catern>
not sure what you mean by "does too"... :)
<catern>
"nix-store --query --hash built-derivation" though I'm not sure it's right
hiratara has joined #nixos
bert_ has quit [(Ping timeout: 250 seconds)]
i-am-the-slime has joined #nixos
<sphalerite>
catern: nix-build --hash builds a fixed-output derivation and places it in the store at the appropriate path, ignoring the outputHash (sha256 in fetchurl terminology) and outputting the correct one
hamishmack has joined #nixos
<sphalerite>
hm, tab completion on /nix/store/foo is getting really slow for me
Capprentice has quit [(Quit: Leaving)]
whatnoise has quit [(Ping timeout: 260 seconds)]
jsgrant has joined #nixos
hiratara has quit [(Read error: Connection reset by peer)]
coot_ has quit [(Quit: coot_)]
dbmikus__ has quit [(Ping timeout: 248 seconds)]
zzamboni has quit [(Quit: Leaving.)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
hiratara has joined #nixos
Lisanna has joined #nixos
cybrian has joined #nixos
mk-fg has quit [(Ping timeout: 240 seconds)]
mk-fg has joined #nixos
mk-fg has quit [(Changing host)]
mk-fg has joined #nixos
Synthetica has joined #nixos
jensens has joined #nixos
thblt has quit [(Ping timeout: 240 seconds)]
cement_ has quit [(Ping timeout: 255 seconds)]
<catern>
sphalerite: yes, but it does it inside the Nix sandbox
<sphalerite>
ooh right
civodul has quit [(Remote host closed the connection)]
<catern>
nix-build --hash does sound useful, but I really really need to be able to run unsandboxed fixed-output builds
<catern>
I think such an ability can form the foundation of some pretty interesting things :)
zzamboni has joined #nixos
cybrian has quit [(Read error: Connection reset by peer)]
i-am-the-slime has quit [(Ping timeout: 255 seconds)]
Neo-- has quit [(Ping timeout: 240 seconds)]
drakonis has joined #nixos
astronavt has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 2 new commits to master: https://git.io/vbESp
<NixOS_GitHub>
nixpkgs/master 8b358b8 Tuomas Tynkkynen: U-Boot: Add a patch to increase max kernel command line length
<NixOS_GitHub>
nixpkgs/master bedc18c Tuomas Tynkkynen: U-Boot: Add ubootQemuArm
<samueldr>
I'm using a device with an Atom x5-Z8350, and their kernel fork with patches works really good*
<samueldr>
maybe it'll be helpful for the x7-z8750 in the gpd pocket
<andi->
I am really just looking forward to a mobile device to use neomutt on ;-)
<andi->
since smartphone companies decided to remove proper keyboards..
<samueldr>
if it's built on the same kind of reference design as the hardware I'm using, there are/were issues with upstream linux
* andi-
still misses his Nokia N900
<samueldr>
my cheap tablet/notebook wouldn't charge with the upstream kernel
<andi->
I read on reddit that someone had it working with upstream kernel (4.14.4) without issues besides some screen brightness (probably wrong x driver, etc..)
<samueldr>
(some of the work generally done by firmwares is passed to the OS with that platform)
<Ralith>
andi-: it doesn't have a cellular modem, though, does it?
dalaing has left #nixos ["User left"]
<andi->
nope :/
<sphalerite>
andi-: fwiw I'm planning to get a librem 5 and get nixos running on that when it comes out in Jan 2019
<samueldr>
:D
<samueldr>
pre-ordered?
<sphalerite>
Shame it's still quite a long way off... But I'll wait!
<samueldr>
yeah, that waiting time is the only reason I haven't pre-ordered / pledged :(
<sphalerite>
No, I figured I'd graduate and get a job and earn some money first
<samueldr>
:)
<sphalerite>
Otherwise I'd have pledged
<samueldr>
late next year I too will check to get one ordered; I'll race ya :)
<sphalerite>
OTT, if anyone's looking for a nixos fanboy (fanperson) to employ next year... I'll be looking for something starting sometime in Q3! (I guess it's still a bit early just now, but just in case :D )
mkoenig_ has quit [(Remote host closed the connection)]