<colemickens>
which version of the librem 13 do you have?
<monokrome>
Yeah, I ordered the v4 model but I think the v3 arrived w v4 hardware
<samueldr>
apparently 2,3 and 4 have the thing in the hardware db
<colemickens>
oops, good find samueldr
<monokrome>
colemickens: I'm not 100% sure - I ordered a v4 but it arrived w/ a case that says v3
<monokrome>
but it has an intel i7 7500U
<samueldr>
monokrome: are you us-based expecting a 104-key like (sometimes called ANSI) layout, and using a us qwerty layout on the device?
<monokrome>
and the site said everything was i5
<monokrome>
so something is fked
<colemickens>
lol that would drive me nuts
<monokrome>
samueldr: Maybe?
<colemickens>
Maybe it's possible you have a weird/new device ID that doesn't match one of the existing entries?
<monokrome>
colemickens: It also came w/ a stripped screw, so getting this NVMe drive in was painful
<monokrome>
lol
<monokrome>
colemickens: Yeah, I'm guessing so
<samueldr>
the 104 (ansi) layout has the enter key "one key high", while the 105 (sometimes called ISO) has the enter key "two keys high"
<monokrome>
but I don't know how to get it to know to treat mine like that
jpaus has joined #nixos
<monokrome>
samueldr: but it turns forwardslash into <
<monokrome>
nothing to do w enter?
<monokrome>
and pipe is >
<samueldr>
yeah, it's "complicated"
<monokrome>
hmm
<samueldr>
see, keyboards are complicated :)
<monokrome>
So I should try a different keymap?
<samueldr>
what's written on the key isn't what's sent to the computer, it's a keycode
m0rphism has quit [Ping timeout: 246 seconds]
<samueldr>
so saying "pipe is >" means nothing most of the time... BUT seeing there is something in systemd for that makes me think there is something behind it
<samueldr>
let me look for something
<samueldr>
meanwhile monokrome, run cat /sys/devices/virtual/dmi/id/product_{family,name}
<samueldr>
I think it should print interesting information related to the version
<monokrome>
I just meant that the explanation you gave maybe doesn't make sense because then it'd press enter, not < - right?
<samueldr>
well not really, what happens is that on some layout the keycode for the key in the location you're thinking of is switched with another one
<monokrome>
Librem 13"
<samueldr>
monokrome: that's the output from the command?
<monokrome>
"Librem 13 v4"
<monokrome>
ya
<monokrome>
Librem 13
<monokrome>
Librem 13 v4
<monokrome>
two lines
<samueldr>
good
<samueldr>
then it should match the v4 things
<monokrome>
So, super funny that the case says "version 3" and that said v4
<samueldr>
looks like the issue is "related" to what I had in mind, in that the workaround breaks keymaps for other kind of layouts :/ (see previous links)
Fare has joined #nixos
<samueldr>
I'd hate to have to deal with that it looks like they're in a weird situation
<samueldr>
someone with a 105 key (iso) layout would need that workaround to be disabled, AFAICT
Makaveli7 has quit [Ping timeout: 246 seconds]
<haslersn[m]>
Is `nixpkgs/lib` documented?
erictapen has quit [Ping timeout: 246 seconds]
<samueldr>
yes, but not entirely, but progress has been made; I think some of it requires unstable/19.03 so the nixpkgs manual on the website might not have everything that's been documented
astrofog has joined #nixos
m0rphism has joined #nixos
<haslersn[m]>
Including the module system?
vincentpants has quit [Ping timeout: 246 seconds]
<samueldr>
no, this is in the nixos manual, not documented in the same way
<aanderse>
under the "additional config options" i would make specific mention of options which can have no default value (like the davmail url, for example), and specifically mention how it is important to keep things like that as regular nixos, generally speaking
astrofog has quit [Remote host closed the connection]
<aanderse>
going back to that whole "as close to service.foobar.enable = true; and you're up and running" philosophy
<haslersn[m]>
infinisil: I want type check of a nested attribute set with certain attributes that can or must appear, so it's a clear fit to the module system.
<aanderse>
i'd also make special mention of options relating to credentials
<aanderse>
i'd hate to see a bunch of new modules come in with people shoving passwords into them
<infinisil>
aanderse: Ah yeah
<infinisil>
aanderse: Although regarding the password thing: If the config file has a password option, then it's only natural that that works too by default
<samueldr>
all talks of RFC, but no link
<aanderse>
lol
<infinisil>
aanderse: I guess I could make it a sidenote along the lines of: "If the config file has a password option, consider using a more secure approach instead, by e.g. using a placeholder password and replacing it at runtime with the actual one"
<aanderse>
yeah
<aanderse>
i think generally database options should remain nixos options
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjJCB
marusich has joined #nixos
random_yanek has quit [Ping timeout: 245 seconds]
<samueldr>
aanderse: could you use `strace -fo davmail.strace $(nix-build --no-out-link channel:nixos-unstable -A davmail)/bin/davmail` to have an strace, and check if around the end there is usage of gdk_pixbuf 2.36 and 2.38 in the file?
<samueldr>
it apparently doesn't crash on my system, which is weird since I had issues beforehand with gdk things, but I just lately did a huge cleanup of my system and might have removed something that littered a gdk_pixbuf loader
* samueldr
is now curious as to why it doesn't fail anymore
<samueldr>
basically, gdk-pixbuf (- instead of _) 2.38 is loading a 2.36 component, which should have been fine, and they are ABI compatible, but a change in libpng (iirc) made them incompatible :/
<{^_^}>
#54278 (by minijackson, 9 weeks ago, open): Programs using gdk_pixbuf 2.38 crash with 2.36 loaders from system
o1lo01ol1o has quit [Remote host closed the connection]
<aanderse>
oh... shouldn't that impact a huge number of gtk based applications?
<samueldr>
yes
<samueldr>
I had made a list of everything that could be potentially failing to work when mixing
<samueldr>
(should be in the issue)
<aanderse>
ok great now i know :)
<aanderse>
thanks
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjJCr
<aanderse>
ok i'm out, have a nice evening everyone
<samueldr>
'night to you!
<aanderse>
thanks for pointing out that issue btw :)
<haslersn[m]>
aanderse: Can I specify that a submodule can have arbitrary keys?
drakonis has joined #nixos
<infinisil>
haslersn[m]: `attrsOf ...` doesn't work for you?
<haslersn[m]>
Not sure. I want to specify that a certain name must appear with a certain value type, but arbitrary other names are allowed to appear.
init_6 has joined #nixos
illegalprime has quit [Ping timeout: 272 seconds]
<infinisil>
haslersn[m]: There's nothing builtin for that, but you can use `attrsOf ...` and then add an `assertions = [{ assertion = cfg.option ? some-key; message = "option needs some-key"; }];`
lewo has quit [Remote host closed the connection]
hio has joined #nixos
ninjin has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjJWq
o1lo01ol1o has quit [Ping timeout: 255 seconds]
<haslersn[m]>
Is there a `nixpkgs/lib` function to flip the order of two keys in a nested attribute set?
<gchristensen>
keys are ordered alphabetically
<gchristensen>
or am I misunderstanding?
<haslersn[m]>
Before: a.b = c
<haslersn[m]>
After: b.a = c
<gchristensen>
{ a = { b = [ 1 2 3 ]; c = [ 4 5 6 ]; }; d = [ 7 8 9 ]; } <- what here?
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<haslersn[m]>
That's already prevented by my type checking, so an error (that would never occur) is fine. Of course not every algorithm can work on every import
<haslersn[m]>
*on every input
hamishmack has joined #nixos
<gchristensen>
aye :)
jpaus has joined #nixos
<gchristensen>
I doubt it
Supersonic112 has joined #nixos
Supersonic has quit [Disconnected by services]
silver has quit [Read error: Connection reset by peer]
Supersonic112 is now known as Supersonic
phreedom_ has quit [Remote host closed the connection]
phreedom has joined #nixos
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @biggs to master « doc/python: remove unnecessary let statement »: https://git.io/fjJW3
<haslersn[m]>
infinisil: I have multiple hosts and some configuration options per host, particularly something like `<hostname>.extensions.<extensionName><instanceName> = {...}`. Another part of the code must be called once per instance but with all hosts that participate in that instance, so I thought I just strip out the `extensions.<extensionName>` and then flip the remaining two keys (`<hostname>.<instanceName>` to
<haslersn[m]>
`<instanceName>.<hostName>`).
<infinisil>
haslersn[m]: And there can only ever be one host for any instance of any extension?
<haslersn[m]>
No, but the "flip" can still be done
<haslersn[m]>
Because after specifying both, it's unique again, since it was before
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<gchristensen>
can you provide several examples of varying complexity?
<gchristensen>
of literal Nix expressions, in the format of `assert (flip foo) == bar;`
endformationage has quit [Ping timeout: 250 seconds]
iyzsong has joined #nixos
bemeurer has quit [Quit: WeeChat 2.4]
linarcx has joined #nixos
marler8997 has joined #nixos
Fare has quit [Ping timeout: 246 seconds]
<marler8997>
Hey, I'm going to see if I can fix https://github.com/NixOS/nixpkgs/issues/58070 I just need a bit of info....I'm going to need to play around with the kernel config, but I don't know how to build the nixos minimal iso image...is there a guide for that somewhere?
<{^_^}>
#58070 (by marler8997, 1 day ago, open): VirtualBox display fails when `VMSVGA` Graphics Controller
<linarcx>
Gyus, when i wanto to build a package locally, i get an error: `called without required argument 'qmake', at /home/linarcx/nixpkgs/lib/customisation.nix:69:12`. why?
jpaus has joined #nixos
palo1 has joined #nixos
palo has quit [Ping timeout: 246 seconds]
palo1 is now known as palo
jpaus has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @joachifm merged pull request #57449 → samsung-unified-linux-driver_1_00_36, tested on nixos x86_64 with m2070w printer/scanner, fixes https://github.com/NixOS… → https://git.io/fhjRD
hyper_ch has quit [Read error: Connection reset by peer]
vincentpants has joined #nixos
[Leary] has joined #nixos
<slack1256>
I was updating my system when I saw it was downloading a path which read "coinMP". I didn't know what closure was pulling it in, so I launched nix-store --query --referrers and got libreoffice as the culprit
<slack1256>
I searched coinMP on duckduckgo and the first result is for a project called CoinIMP which is a "free javascript browser miner".
Lears has quit [Read error: No route to host]
<slack1256>
I thought there was a bad actor on nixpkgs bundling a javascript miner on libreoffice, so I went to the libreoffice dir and searched for coinMP.
NightTrain has quit [Quit: Lost terminal]
<slack1256>
Results there is a second project called COIN-OR which makes lineal/convex programming software, and libreoffice calc uses a engine called coinMP (mixed integer programming?)
<{^_^}>
[nixpkgs] @andir pushed 4 commits to release-19.03: https://git.io/fjJ88
<andi->
tilpner: I guess we could/should add a `unique` after merging all the options in the `listOf` type... Mind opening a PR? :-)
marusich has quit [Remote host closed the connection]
vincentpants has quit [Ping timeout: 246 seconds]
agander has joined #nixos
hakujin has joined #nixos
<etu>
Hmmmh, I have a system that has a mountpoint to automount a NFS share. And it's set to "noauto,x-systemd.automount" and so on
<etu>
But when I boot it, kodi manages to start before the dhcp is done and kodi touches this folder so it fails to mount 5 times before the network is up
<etu>
So I have to ssh in and mount it -- each reboot.
<etu>
Which isn't often but needs to be remembered. Because the unit has given up...
<etu>
Can I modify the unit (for the mountpoint) to wait a bit between retries or require it to be after the network is actually up or something?
<{^_^}>
[nixpkgs] @FRidh pushed 2 commits to release-18.09: https://git.io/fjJ8R
<immae>
Hello there! I have a src path that I build of the form `src = "${./projects}/${some_variable}/";`, and several derivations with different values of "some_variable". However, the way I wrote it makes the derivation dependent on the whole ./projects path instead of just the subfolder, can I optimize it a bit to avoid rebuilding all the derivations each time I modify one subfolder?
jpaus has quit [Ping timeout: 246 seconds]
<tilpner>
immae: Can you do "${./projects + some_variable}"?
<tilpner>
> /foo + /bar
<{^_^}>
/foo/bar
<immae>
I expect to have the same result, but I’ll try it
<tilpner>
No, that should only import /foo/bar inot the store
<andi->
tilpner: ^ I opened a PR for what you said :-)
<tilpner>
andi-: Oh, thanks! I was going to, but wasn't sure how to test it (I try to avoid building my system from a manual checkout)
<tilpner>
andi-++
<{^_^}>
andi-'s karma got increased to 9
<andi->
tilpner: there is nixos tests for that!
<andi->
I should make a new test for that specific case. Feel free to push into my branch / comment on the PR if you want to add something
<tilpner>
andi-: While you're at it, it should probably be with types; uniq (listOf port)
<andi->
yeah, will get to that next. Currently looking into the report that our systemd doesn't build anymore because the patches we fetch vanished :/
<tilpner>
Ha, it even has a comment explaining why the patches might vanish :D
<andi->
I will probably change the source line to something on snapshots.debian.org since that is supposed to never go away..
<andi->
We want reproducibility even in a few years, right?
<ircuser5678>
Mh, somehow I can't find out why openssh refuses my (damn) key. Even when looking at the log (running at DEBUG3 level)... I've tried a 4096 bit RSA-2 key (generate by PuTTY Key Generator) and a 2048 bit one just to make sure...
<tilpner>
andi-: Shouldn't this be covered by tarballs.nixos.org?
<tilpner>
Oh, maybe not
<andi->
tilpner: I think the tarballs.nixos.org thing has to be updated semi-manually? Not sure how that works
<tilpner>
I don't know either, I was assuming it to be automatic for everything, but then I read the index page
linarcx has quit [Ping timeout: 250 seconds]
<tilpner>
It's probably a good idea to switch to stable tarball locations anyway
linarcx has joined #nixos
<{^_^}>
[nixpkgs] @andir opened pull request #58168 → systemd: update debian patches url to snapshots.debian.org → https://git.io/fjJ8b
<inquisitiv3>
I'm trying to install home-manager on my Fedora system with Nix. Step 2 make me a little unsure on what I should do. I've run the `nix-channel` commands, and home-manager shows up when I run `nix-channel --list`. But should I change the $NIX_PATH? The wording in the instructions make it sound that it isn't everytime you need to do that on a system with just Nix.
<inquisitiv3>
And if I need to do that, should I change the .bashrc or my fish shell config? The linked issue is a little over my head.
<andi->
inquisitiv3: try #home-manager if you do not get a response here. There is a channel for it now :-)
<andi->
tilpner: we could test the firewall settings on module level. No need to actually touch the VM tests.. I just fail to remember/find a place were we have such tests :/
<tilpner>
There are other tests than <nixpkgs/nixos/tests>?
<andi->
I think there are a ew build-support tests in nixpkgs but there surely must be something for all the module magic?!?
ircuser5678 has quit [Quit: Going offline, see ya! (www.adiirc.com)]
alex`` has joined #nixos
<tilpner>
<nixpkgs/lib/tests/modules>
<tilpner>
But that's just for the module system, not the NixOS modules
<andi->
We also probably do not want an immediate abort of the evaluation just because some assertion failed.. Maybe a VM test is best after all... It just feels a bit heavy for such a change.
<tilpner>
andi-: Maybe not. The VM test could ask iptables how many rules were created for each port, so that would test more than just the module system
<andi->
yeah
<andi->
tilpner: well `uniq` is not intended to be defined in more then one location.. So maybe I have to adjust my approach
<andi->
it doesn't mean the values are unique..
<tilpner>
So just listOf ports and lib.unique later?
<asthma[m]>
is it possible to use nix / nixops to create a deployment of a postgres db and a couple of webserver? I can do it with Docker, but I'd like to use nix
jasongrossman has quit [Ping timeout: 268 seconds]
<noonien>
any reason why this config (http://vpaste.net/GrsFK) on a rPI 3B+, does not have `sshd`, and `wpa_supplicant` in `systemctl list-units --all`, while both `wpa_supplicant.service` (http://vpaste.net/t73eT) and `sshd.service`(http://vpaste.net/MSajp) exist in /run/current-system/etc/systemd/system/?
jasongrossman has joined #nixos
jackdk_ has joined #nixos
achw1 has joined #nixos
Rusty1 has joined #nixos
<noonien>
(this was after a reboot, and systemctl daemon-reload)
Makaveli7 has quit [Ping timeout: 245 seconds]
achw1 has quit [Read error: Connection reset by peer]
Makaveli7 has joined #nixos
achw1 has joined #nixos
hakujin has joined #nixos
b has joined #nixos
hio has quit [Quit: Connection closed for inactivity]
<srhb>
noonien: Does systemctl start wpa_supplicant also fail_
iyzsong has joined #nixos
<srhb>
?
<noonien>
hmm, it does not
<noonien>
it starts wpa_supplicant
hakujin has quit [Ping timeout: 255 seconds]
<srhb>
noonien: Then the answer, as it was yesterday, is that nixos/modules/profiles/installation-device.nix removes the wantedBy from the units.
<noonien>
sure, but why doesn't openssh start either?
<srhb>
Same reason.
<noonien>
OH
<noonien>
they're both on the same line
<noonien>
my bad
<srhb>
Sorry, I formatted it like that for irc :)
<noonien>
oops, yeah, my bad
<srhb>
So you'll want to either not reuse that installation import, or reintroduce a dependency upon "multi-user.target" or something like that
<noonien>
yeah, should have pursuied that yesterday, i've seen that line in instllation-device, but missed the ssh one, thought there might have been a different issue
<noonien>
thanks!
<noonien>
srhb++
<{^_^}>
srhb's karma got increased to 38
<srhb>
Sure thing.
hio has joined #nixos
rprije has quit [Ping timeout: 246 seconds]
zupo has joined #nixos
ym555 has joined #nixos
<betawaffle>
does anyone use 1password on nixos?
vasarmilan has joined #nixos
<vasarmilan>
hi! I'm trying to make a nixos-container have network access. I've tried copy-pasting the example from here: https://nixos.org/nixos/manual/#ch-containers only with replacing "eth0" with the interface of my wifi card; however from inside the container pinging a website returns "Name or service not known". also I cannot set up bridging from a wireless network.
<vasarmilan>
anyone has an idea how I can let a container access the internet?
<vasarmilan>
(with privateNetwork=true)
<vasarmilan>
thank you!
jasongrossman has quit [Read error: Connection reset by peer]
jackdk_ has quit [Remote host closed the connection]
agander_ has joined #nixos
<Uma>
Thanks srhb :)
<alj[m]>
Hello! I looked into cluster Filesystems to put onto my two identical nixos servers. This question was (kinda) asked before in here but can anyone tell me if ocfs2 on top of DRBD is possible on nixos?
agander has quit [Ping timeout: 250 seconds]
<srhb>
Uma: Welcome. Sorry I don't feel confident enough to just click the button, but I really know nothing about it. Feel free to solicit a review from someone else who's touched the commit history of opam if you want in the meantime :)
vasarmilan has quit [Ping timeout: 256 seconds]
iyzsong has quit [Ping timeout: 258 seconds]
jlslnx has joined #nixos
<{^_^}>
[nixpkgs] @alexarice opened pull request #58173 → brillo: init at 1.4.3 → https://git.io/fjJEs
vasarmilan has joined #nixos
ng0 has quit [Remote host closed the connection]
ym555 has quit [Ping timeout: 246 seconds]
ng0 has joined #nixos
jlslnx has quit [Quit: leaving]
freeman42x has quit [Ping timeout: 268 seconds]
vasarmilan has quit [Ping timeout: 256 seconds]
erictapen has quit [Ping timeout: 246 seconds]
<marler8997>
hey, my account on discourse.nixos.org is on hold...I'm a new user...is it normal for that to happen?
<werner291>
Question: do you guys know how to get Dovecot to accept plaintext passwords? I have a user who is stuck on an old email client that doesn't seem to like STARTTLS
<werner291>
Logging in works perfectly on Thunderbird as well as the iPhone email client.
<gchristensen>
oh dear
<infinisil>
ward: What's this got to do with NixOS?
<infinisil>
werner291: ^
<werner291>
Because the email server is hosted on NixOS
<gchristensen>
werner291: it is very, very dangerous to accept plaintext passwords
<gchristensen>
I'd recommend opening an issue on that repo and asking for the help there -- they'll be able to help more
<samueldr>
ggpeti: the evaluation failure was fixed quite recently, and the last eval apparently failed "out of the blue" on the build machine
<werner291>
Allright, thanks.
<samueldr>
(post fix)
<MichaelRaskin>
I think old email clients normally supported a separate encrypted port. Although maybe it only supports TLS versions that are insecure.
<werner291>
It's Outlook for Mac from 2011. The user is stuck on that for some reason.
<ggpeti>
samueldr: I only see grey X-s for the past 3 days, what fix are you referring to?
<samueldr>
different failures happening each after the others
<werner291>
The only error message it gives is an orange dot next to the account.
<gchristensen>
werner291: my experience has been that if you tell the user that doing this means anyone can trivially read their email, they're more inclined to upgrade
<samueldr>
(not each one, but a couple per)
<samueldr>
the penultimate one failed due to the anbox derivation not eval'ing for i686-linux, but this was fixed in af4ad24d73d203618a177381027355e0b4a57c66
<samueldr>
I restarted the tarball build to see if it was transient
<ggpeti>
oh, good. let's hope
<werner291>
They're aware of the risks.
<samueldr>
(verifying locally for luck)
<gchristensen>
ok
endformationage has joined #nixos
<jasongrossman>
werner291: I expect this conversation is annoying for you, because the user might not want to know, but ... have you asked them to try MailMate? It looks very similar to Apple Mail (unlike e.g. Thunderbird).
<jasongrossman>
werner291: Oh, sorry, you said Outlook. For some reason I read that as Mail.
<werner291>
They're willing to switch, but they seem to have a shitton of email accounts.
<jasongrossman>
Well MailMate is very good, FWIW.
<werner291>
They've been building up their email situation over about 10 years, it's gonna take a lot of work to switch.
<werner291>
Sigh... What have I gotten myself into?
<jasongrossman>
werner291: I sympathise. Might be best to look for Dovecot forums.
<gchristensen>
I charge hazard pay for email administration :P
alex`` has quit [Quit: WeeChat 2.4]
<ggpeti>
samueldr: is there a way to estimate when a queued build will run?
<werner291>
They are willing to switch to Thunderbird if I get their other accounts working, but oh well... Thanks for the sympathy, I'll keep looking.
<samueldr>
toss a couple of coins and guess? (sorry, I really don't know)
<jasongrossman>
werner291: I expect you know this, but there is a standalone dovecot package for NixOS (in addition to simple-nixos-mailserver).
<jasongrossman>
infinisil: Yes, I know. So there. :P
<werner291>
Found it... You can't turn on plain auth without turning off ssl completely, but the user just suggested a workaround.
<jasongrossman>
werner291: \o/!
<werner291>
To redirect the mail to one of their other accounts, and not deal with IMAP. They've been complainig about this email client for years, but they seem unwilling to switch.
<werner291>
Just... I'mma go do this, then it's not my problem anymore. Thanks for the help.
<jasongrossman>
werner291: I'm glad the problem's gone.
<samueldr>
my following explanation might not be "right" 100%, but should give the gist of it, ggpeti, an eval is a special kind of build only meant to queue builds from the configured thing to evaluate
<noonien>
infinisil: i have not, how/when are those used?
<ggpeti>
so the evaluations generate the jobs, and the builds actually execute the jobs?
<samueldr>
kind of yeah, a jobset is evaluated, its jobs are built
<ggpeti>
and then the job "unstable" of jobset "trunk" of project "nixpkgs" triggers a release to the channel nixpkgs-unstable
<srhb>
ggpeti: Technically a cron job checks whether there's a new successful build of that (and that the corresponding eval is complete) but yes.
<samueldr>
the project nixpkgs has the jobset trunk composed of one job per package*platform; when eval'd it they are eventually queued as builds.
<samueldr>
there is no "unstable" job (I think?). Hydra has a channels facility, but for Nixpkgs and Nixos it isn't used and, as srhb said, there is another mechanism in place that creates the channel
<samueldr>
I'm not 100% sure why the channels made by Hydra aren't used, but I'd bet that it overloads Hydra with the amount of stuff it already handles
<samueldr>
so, to finish the explanation, nixpkgs-unstable has 1906 queued jobs, of those most are darwin jobs
<samueldr>
until they are all built (or failed) it won't advance
<samueldr>
due to the problematic nature of getting Darwin builders, you might have a better turnaround for channel updates on nixos-unstable instead of nixpkgs-unstable, if you're not targetting Darwin :/
<srhb>
There's a ton of aborted darwin jobs, too... Perhaps we should retrigger those soon.
<ggpeti>
I'm more interested in the setup now than turnaround time
<ggpeti>
how are jobs getting aborted?
<srhb>
ggpeti: Either manually or some specific, repeated build failures
<srhb>
ggpeti: (Like unable to connect to a builder repeatedly)
<Yaniel>
because if that is set it should skip that script entirely
<Yaniel>
but instead I get a build error unless I set that environment variable manually
<Yaniel>
and no stdout from the script to debug it
<clever>
Yaniel: check every one of the buildInputs for a nix-support/setup-hook file
<slack1256>
running `groupmems -l -g users` doesn't show my main user and neither /etc/groups. Yet running `id` says my main GID is users, are those commands for extra-groups only?
<clever>
slack1256: i think users is your default group (as specified by passwd), but your not also a member of it
sinner has joined #nixos
sinner is now known as Guest11136
<slack1256>
So you aren't part of your default group? weird...
Guest3456 has quit [Ping timeout: 250 seconds]
rycwo has joined #nixos
Makaveli7 has quit [Remote host closed the connection]
<marler8997>
how do I map an "attribute name" to the name that would go in systempackages in configuration.nix?
<srhb>
marler8997: Directly.
linarcx has joined #nixos
<srhb>
marler8997: Attribute names as defined in pkgs/top-level/all-packages.nix are pkgs.attrname. As in, pkgs.hello etc.
<marler8997>
right
<marler8997>
but if I have an attribute name
<marler8997>
how do I find out what the corresponding "package name" is?
<samueldr>
> hello.name
<{^_^}>
"hello-2.10"
<srhb>
marler8997: nix eval nixpkgs.hello.name
<srhb>
marler8997: (Though usually, you neither need or want the "package name" for anything at all)
<samueldr>
though environment.systemPackages don't use the name
<marler8997>
oh...systemPackages uses the attribute name?
<srhb>
Yep. Everything does, aside from nix-env version checks
<srhb>
The name is irrelevant in most respects.
<marler8997>
ah..very good
<srhb>
That* name
<samueldr>
the nix store path is the one other main user
periklis has joined #nixos
<marler8997>
thanks for the help...there's alot to learn...going through nix pills for the second time and still trying to grasp all these new concepts
<srhb>
marler8997: Yeah, lots indeed. It's a process :-)
<noonien>
what's an easy way of getting the current running system's configuration.nix?
<noonien>
because i've got the impression, that even after an install, the configuration is not properly applied
Uma has quit [Quit: leaving]
Fare has joined #nixos
<fgaz>
Is there a way to create a nixos container with a specific nixpkgs checkout?
<fgaz>
Is there a way to create a nixos container with a specific nixpkgs checkout?
<samueldr>
noonien: might be able to figure things out using `nixos-option`
o1lo01ol1o has quit [Remote host closed the connection]
<samueldr>
here it prints where it's declared, where it's defined, and the value + default
<fgaz>
There's no -I and nixpkgs.pkgs does not affect the modules
<samueldr>
though I'm not sure whether the value comes from the just-interpreted config or from the running config
<slack1256>
fgaz: You could checkout nixpkgs at the checkout and pass as an argument that repo as the one from where to build from
<slack1256>
nix-{build,env,shell} all take a repo as a argument, I guess container can be made the same wway
<fgaz>
slack1256 nope, nixos-container takes surprisingly few options
<fgaz>
Oh wait there's the NIX_PATH env variable though
<fgaz>
Lemme try
<noonien>
full story, i've started with http://vpaste.net/GrsFK, i now have: http://vpaste.net/E4BjE however, after switch + reboot, i still get root autologin (which i've not enabled in my config), this is pretty weird
<noonien>
nixos-option shows the right values
<noonien>
`man` also works
<samueldr>
after switch + reboot?
<noonien>
yes
<samueldr>
this is likely that the bootloader isn't loading the right generation
<fgaz>
It works! I had to prevent sudo from rewriting the env though
<samueldr>
noonien: this looks like ARM, and u-boot
<noonien>
i've not interacted with the bootloader, which selected the default boot options
<samueldr>
noonien: are you trying to stop using the /boot FAT partition to hold the boot files?
<noonien>
yes, this is on a rPI 3B+
<noonien>
nope, i just want to customize my config
<noonien>
don't have any specific requirements other than this
<samueldr>
okay, and yeah, looks like you're still mounting /boot
<noonien>
in the past i was not able to get wpa_supplicant and sshd to start on boot, even by setting their "wantedBy" systemd service option
Nycatelos has left #nixos ["WeeChat 2.4"]
<noonien>
i had hoped compeltely configuring everything might solve that, however, switch+reboot seems to load a weird version of the system
<noonien>
i'll do another switch, reboot, collect garbage, reboot
<samueldr>
hmm, weird, short of the /boot partition not being updated, only by looking, it looks right
<noonien>
just to try to have minimal variables
<samueldr>
before rebooting
<samueldr>
copy the output of nixos-rebuild switch
HappyEnt has quit [Remote host closed the connection]
HappyEnt has joined #nixos
cyraxjoe has quit [Ping timeout: 272 seconds]
MightyJoe has joined #nixos
o1lo01ol1o has joined #nixos
<jD91mZM2>
Hey, I'm finally learning NixOps! I love it to bits, it's so cool to see it working. One thing though, I have created a new user in hope to be able to connect to it via SSH, but `nixops -d my-deployment my-user@my-machine` asks for a password. Is that supposed to happen until I configure a SSH key? For details, this is a DigitalOcean deployment with a (yet) very small config:
<tilpner>
jD91mZM2: I don't use nixops, but I would just set users.users.user.openssh.authorizedKeys.keyFiles = ./jdsomething.pub;
ggpeti has quit [Ping timeout: 256 seconds]
<icey_>
I've been trying to get nixos set up and I'm struggling with overlays a bit. I sent a PR adding a setting to grub which hasn't yet made it into master, so I wanted to use an overlay to override grub on my system with the new version from my fork, keeping all the rest of packages from upstream
<icey_>
somehow, I can't seem to get the overlays get picked up
<tilpner>
"This can either be the full path to the module or a string with the filename relative to the modules path (eg. <nixpkgs/nixos/modules> for nixos)."
krig has joined #nixos
strexicious has joined #nixos
<tilpner>
Because this is so clumsy, many people build their systems directly from their nixpkgs fork, to which they can make changes at will and without this replacement dance
<icey_>
thx, I'll study it!
<icey_>
well, I'm in the process of learning nixos, so may just as well carry out the exercise once ;-)
<strexicious>
hello, I am trying to install nixos from minimal installer, but wifi adapter doesn't seem to work. dmesg says it failed to read from the device, rfkill showed it's softblocked. the adapter is atheros.
fragamus has joined #nixos
drakonis has quit [Ping timeout: 250 seconds]
<Rusty1>
try: sudo rfkill unblock all
<strexicious>
yeah, did rfkill unblock 0
<strexicious>
and it unblocked
<strexicious>
but ifconfig is still not showing wlan0 or any wireless interface
<strexicious>
tried modprobe ath9k as well
alexarice[m] has joined #nixos
mmlb977 has joined #nixos
<MichaelRaskin>
No wlp… ?
<strexicious>
nope
<strexicious>
I am using lenovo ideapad 330
o1lo01ol1o has quit [Read error: Connection reset by peer]
sbodin has joined #nixos
<jD91mZM2>
tilpner: Sorry for the late response, I had dinner. That option has a warning about NixOps which makes me think that it is supposed to allow public key logins by default, or at least some other way.
abathur has joined #nixos
drakonis has joined #nixos
dnlkrgr has quit [Ping timeout: 246 seconds]
<jD91mZM2>
Sorry for the ping, gchristensen, but I think I'll need to summon you here to read my question above. It's quite large, so I won't post it again, but you won't miss it.
<tilpner>
jD91mZM2: I vaguely remember nixops creates its own ssh keys, but that didn't fit my usecase at all
<tilpner>
jD91mZM2: "Note that we rely on a ssh key resource with the hard-coded name ssh-key. Providing your own key is not supported yet." seems relevant
<jD91mZM2>
Yes, I'm using the default. It allows me to login to root via `nixops ssh ...`. But the question is how to make that key also work for another user
<jD91mZM2>
(tilpner)
* jD91mZM2
has to go
<tilpner>
Oh. I don't think I can help you
<{^_^}>
[nixpkgs] @PombeirP opened pull request #58182 → conan: fix pluginbase version requirement → https://git.io/fjJ2H
ddellacosta has joined #nixos
lassulus has quit [Ping timeout: 246 seconds]
alexarice[m] is now known as AlexRice[m]
agander_ has joined #nixos
ggpeti has joined #nixos
ddellacosta has quit [Ping timeout: 246 seconds]
ddellacosta has joined #nixos
agander has quit [Ping timeout: 246 seconds]
drakonis1 has joined #nixos
fragamus_ has joined #nixos
fragamus has quit [Read error: Connection reset by peer]
o1lo01ol1o has joined #nixos
ggpeti has quit [Ping timeout: 256 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @tohl opened pull request #58183 → added thunar-nilfs2-plugin, tested on nixos x86_64 → https://git.io/fjJaJ
<tilpner>
Look into overlays for an even cleaner setup. They would allow you to use pkgs.now-cli or pkgs.your-custom-now-cli
Guest34205 has quit [Ping timeout: 244 seconds]
<nhill>
ok
<nhill>
haven't yet read the actual language spec for nix, still new
<tilpner>
(\n are newline characters)
<tilpner>
(So you replace them with a new line in the file. But that's not required, just prettier)
<nhill>
tilpner: well, updating the version and sha seemed to work :+1:
<tilpner>
yay
<MichaelRaskin>
That's annoying: I can run Terminator in nix-shell, but making all this double-wrapping work seems nontrivial…
<scott>
does anyone know where to start with debugging Steam problems? i can't run Steam Play / proton games and it's not clear why
<tilpner>
scott: Can't run? What do you mean?
<scott>
the games just never open
<tilpner>
But Steam says Running?
<scott>
i could only find a couple nixpkgs github issues about this and it seems like other people were able to run the games i can't, but i'm unsure how to get more information on why it doesn't work
<scott>
tilpner: for a few moments, then it goes away
<tilpner>
I suggest adding a terminal emulator as a game to Steam
<AlexRice[m]>
which game are you trying?
<tilpner>
Launch that, and then manually launch your game
<scott>
right now i'm trying The Witness
<{^_^}>
[nixpkgs] @NathanielHill opened pull request #58184 → Update now-cli to v14.2.3 → https://git.io/fjJar
<scott>
but all proton games fail
ym555 has joined #nixos
<tilpner>
That should give you more debug output and the ability to strace
<scott>
tilpner: will do
<scott>
manually launch as in `steam steam://<game id>` or something more specific?
<tilpner>
Navigate to the executable and launch it
<scott>
if it's more specific, i don't know how to launch proton games. there's just windows exes in the game dir
<MichaelRaskin>
Yes, Proton is their branch of Wine
<tilpner>
You can observe the parameters Steam uses with ps or htop
<AlexRice[m]>
does steam bundle proton into the executable?
nhill has quit [Quit: leaving]
<scott>
no, i think steam installs each proton version once globally, not per-game
<MichaelRaskin>
I guess you can also attach strace to a running Steam, or does Steam detect that and complain?
<AlexRice[m]>
so navigating to the executable might not work?
<tilpner>
MichaelRaskin: That would make for a noisier output
<tilpner>
MichaelRaskin: And attaching to a running process isn't easy if it just dies immediately
<MichaelRaskin>
That's true, but you can filter by PID post-factum
<MichaelRaskin>
You can attach to Steam UI and request tracing all children
<tilpner>
Sure, it's preference of course
<MichaelRaskin>
I don't use Steam — is it easy to locally run a terminl via Steam?
<scott>
tilpner: hrm, i tried adding konsole and xterm inside steam and it won't run them
<scott>
can i use the `steam-run` thing, or is it a bit different?
wfranzini has quit [Remote host closed the connection]
<tilpner>
Unfortunately yes. I'm unsure of the details, but I think the environment Steam sets is a little different, and some games might behave differently depending on if they can connect to Steam (which only works if it's launched from within Steam AFAICT)
wfranzini has joined #nixos
* tilpner
answers or-question with yes
* tilpner
meant "it's a bit different"
<scott>
ah nevermind, figured out how to make it launch a terminal
<scott>
i can think of one potential weirdness: the steam library i have The Witness installed in is on NTFS and the proton wine prefix is placed on the same filesystem. maybe wine dislike that?
<srhb>
benley: woops :-)
pie_ has joined #nixos
<scott>
i could try installing it to my ext4 instead
erasmus has quit [Ping timeout: 250 seconds]
<benley>
srhb: I'm still pretty thrilled that someone packaged anbox for nixos in the first place :-D
rcshm has joined #nixos
<srhb>
benley: I hope everything is OK with it now, the reason it snuck through is that I was focused on restoring the entire nixpkgs build and didn't pay enough attention to the package. Still, everything _should_ be okay, I think.
<srhb>
nixpkgs tarball build*
<srhb>
Seems to build fine here though :)
<MichaelRaskin>
Fare: I can lunch Terminator in nix-shell, but I officially give up on fixing the package
o1lo01ol1o has quit [Remote host closed the connection]
<noonien>
i'm seeing some zfs-related systemd processes, even though i'm not using zfs, is this normal?
<noonien>
i'm trying to cut down on stuff that runs, since this is an embedded system
<infinisil>
noonien: Like what?
leat2 has joined #nixos
<noonien>
hmm, actually, the normal sd-image seems to have had some ZFS stuff, my bare-bones config doesn't seem to have anything zfs-related in the journal
<{^_^}>
[nixpkgs] @kamidon opened pull request #58187 → password-store: link ext. man pages into the password-store $out → https://git.io/fjJVt
logand has joined #nixos
<eyJhb>
Can anybody recommend a terminal replacement, currently using urxvt, and I feel like I should change... ranger image preview would be awesome
<tilpner>
You might get better recommendations if you specify why you feel you should change
<logand>
eyjhb: st + ranger image preview works
o1lo01ol1o has joined #nixos
<scott>
AlexRice[m]: tilpner: i moved my .steam and .local/share/Steam to a backup folder and let steam re-install, then ran Castle Crashers (which also failed before) and it worked. i don't have time now, but later i'll disambiguate whether it was the Steam reinstall or playing the game from ext4 rather than NTFS that made the difference
<ldlework>
eyjhb: I have ranger image preview working in urxvt...
<eyJhb>
ldlework: it isn't because it isn't working, I just feel that quite often e.g. the clipboard functionality screws me over, or url opening
<benley>
srhb: I got it installed and running, which is promising
<eyJhb>
logand: does it have vim like marking?
<benley>
srhb: I don't really know how to do anything _useful_ with it yet, but it's exciting to have the runtime working
<logand>
eyjhb: it probablu has nothing to do with the terminal, it uses w3m-img to draw the image iirc
vincentpants has quit [Ping timeout: 245 seconds]
<ldlework>
eyjhb: you should use autocutsel regardless of the terminal emulator and urxvt opens urls
<eyJhb>
ldlework:to keep them in-sync?
<ldlework>
yeah, easy to add in nixos too
<eyJhb>
I would imagine just adding it as a package, and then enabling a service?
<eyJhb>
Or starting it in bashrc
<ldlework>
i haven't thought about linux's clipboard situation in a decade
<eyJhb>
But it is only SOMETIMES it screws up... And then opening a new terminal fixes it again
<logand>
LANG=ja_JP.UTF-8 thunar uses Japanese GUI, but LANG=ja_JP.UTF-8 firefox uses english, no matter what LANG i specify (i have them installed and are listed in locale -a). is there a way to specify, which language should firefox use?
<eyJhb>
ldlework: I feel like that isn't a network manager applet ;)
<logand>
tilpner: i have config.i18n.defaultLocale set to english but some users want locale which is different from defaultLocale but is in config.i18n.supportedLocales
periklis has quit [Remote host closed the connection]
Lears has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
<logand>
tilpner: obviously japanese defaultLocale is not good for me, i'd be completelly lost
acowley has quit [Ping timeout: 255 seconds]
[Leary] has quit [Read error: Connection reset by peer]
<tilpner>
logand: Just .override firefox-bin with a different config
acowley has joined #nixos
<logand>
tilpner: i see now, thanks!
<tilpner>
logand: You'll actually want to override firefox-bin-unwrapped, and then wrap it with wrapFirefox. See all-packages.nix for examples
<illegalprime>
somehow setting systemd.services."serial-getty@ttyGS0".enable = true; causes this error don't know how to build these paths (may be caused by read-only store access)
<gchristensen>
where does /dev/fd come from?
<Rusty1>
floopy disks of yore?
<gchristensen>
its a symlink to /proc/self/fd :)
<gchristensen>
I'm playing ultra minimal system games, and am having a hard time finding what create is
<gchristensen>
what creates it*
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
mmlb977 has joined #nixos
<tilpner>
illegalprime: Paste the full error
silver_ has joined #nixos
<illegalprime>
tilpner: i'm very dumb, I had --dry-run on on something I hadn't evaluated yet
<illegalprime>
tilpner: unfortunately it doesn't work, setting `enable = true` causes nix to create a new systemd unit, not use the template unit that already exists
<MichaelRaskin>
I think Xorg just sometimes crashes on suspend for me, not sure if connected to hoteject
mounty has quit [Remote host closed the connection]
mounty has joined #nixos
mekeor has quit [Quit: digital revolution = cryptoanarchism + cybercommunism]
<gchristensen>
MichaelRaskin: are you able to recover, or do you reboot?
<MichaelRaskin>
I just restart X11; screens survive in detached state anyway
<gchristensen>
interesting. how do you do that?
<srhb>
If I attach displays after bootup, I can only get one big display that spans multiple monitors, even though xrandr says that's not what's going on. If I connect them before booting, everything is fine. Go figure.
<MichaelRaskin>
Remember I do not run literally NixOS
<srhb>
multi screen on linux is dark voodoo.
<gchristensen>
i do :)
mmlb9774 has joined #nixos
mmlb977 has quit [Ping timeout: 250 seconds]
<MichaelRaskin>
I login in text console anyway, so I do exactly the same thing I use to start X the first time
alex`` has quit [Ping timeout: 250 seconds]
<gchristensen>
ah, duh, what is your symptom of a crashed X?
<gchristensen>
is it just closed?
<MichaelRaskin>
Yep, _Xorg_ as a process just disappears
shibboleth has joined #nixos
mmlb97745 has joined #nixos
drakonis_ has quit [Remote host closed the connection]
alex`` has joined #nixos
isidore has joined #nixos
mmlb9774 has quit [Ping timeout: 250 seconds]
<MichaelRaskin>
i915+nouveau+modesetting doesn't seem to actually lock up GPU for me
<gchristensen>
I seem to have compiled out support for CAP_* but I'm not sure how :|
<thomashoneyman>
does anyone have a pointer to docs on setting up external monitors?
<haslersn[m]>
We are encouraging people to put built time dependencies into `nativeBuildInputs`, however, I experienced that it does only do what it's supposed to do when also specifying the package as `buildPackages.buildPackages.package-name`. Why is that?
<haslersn[m]>
infinisil: As far as I know, one of the reasons we have things such as `nativeBuildInputs` is to allow cross compilation. But cross compilation works for me only when prefixing the packages in `nativeBuildInputs` with `buildInputs.buildInputs.`. Why is that needed?
Fare has quit [Quit: Leaving]
<srhb>
haslersn[m]: Can you give an example derivation of what you expected to work, and what you had to do instead? It's a bit unclear what's going on right now.
<noonien>
is there a way to add channels from within conviguration.nix?
abathur has joined #nixos
<{^_^}>
[nixpkgs] @veprbl merged pull request #58191 → paulstretch: init at version 2.2-2 → https://git.io/fjJwQ
<{^_^}>
[nixpkgs] @veprbl pushed commit from @Zer0- to master « paulstretch: init at version 2.2-2 »: https://git.io/fjJrn
<srhb>
noonien: The channel mechanism is sort of explicitly about managing the nixpkgs revision in a more or less imperative fashion. There's lots of ways to make a nixpkgs revision available declaratively within your configuration, though
<srhb>
noonien: Like, fetchFromGitHub, say.
<srhb>
noonien: You can also take a look at how the installer sets up a channel if you want.
<noonien>
yes, i would have liked to have channels.nix that represented the current channels i'm working with
<noonien>
and nix-channels would manage that
<noonien>
but it doesn't seem to hard to do myself
<drakonis1>
there's a mechanism getting worked on that would supercede the existing setup
<drakonis1>
would make it all much easier to manage
o1lo01ol1o has quit [Remote host closed the connection]
<noonien>
where can i read more about that?
<drakonis1>
was on nixcon i think
abathur has quit [Ping timeout: 250 seconds]
<drakonis1>
the current form is supposed to be called flakes.nix but i'm not sure what's the status
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<{^_^}>
[nixpkgs] @veprbl pushed commit from @Zer0- to release-19.03 « paulstretch: init at version 2.2-2 »: https://git.io/fjJrC
<srhb>
drakonis1: Is that actually being worked on?
<srhb>
I do hope we got with another name, regardless :P
<drakonis1>
i really just want to get rid of the insane way to manage repositories
<drakonis1>
its currently driving me mad because i don't know how to keep it all up to date in a sane manner
Glider_IRC has quit [Quit: Leaving]
freeman42y has joined #nixos
<noonien>
would channels.nix not be a pretty good solutions?
<noonien>
a file that contains all repos you need, using fetchWhatever
<drakonis1>
probably
<drakonis1>
the nix roadmap vaguely mentioned this
<drakonis1>
speaking of which, it never got posted to the repository issues, has it?
vincentpants has quit [Ping timeout: 250 seconds]
freeman42x has quit [Ping timeout: 258 seconds]
<{^_^}>
[nixpkgs] @Ma27 closed pull request #56345 → nixos/sshd: implement support for `Match` groups → https://git.io/fhbX7
<noonien>
hmm, would the `nix*` commands work if there was no channel available?
<drakonis1>
i think so
<drakonis1>
nix can use the channels added through the configuration
<drakonis1>
a whole lot better and cleaner than the previous setup
Rusty1 has joined #nixos
<noonien>
hmm, added how?
<noonien>
(sorry, i'm actually pretty new to nix)
<clever>
i prefer using nix-channel, because i dont want it updating on its own, and i prefer to have nixpkgs cached
<drakonis1>
i swear i've seen you from somewhere before
<drakonis1>
i'd rather just have the ability to declaratively define nixpkgs as a channel without hacks
<noonien>
hehe, i'm in quite a few channels
o1lo01ol1o has joined #nixos
<drakonis1>
i just override my packages with nixos-unstable
<drakonis1>
works decently well but only with nix
<noonien>
clever: how/why would it update on its own?
vincentpants has joined #nixos
<clever>
noonien: the new style is things like channel:nixos-unstable in NIX_PATH, and that downloads the current nixos-unstable every time the cache expires (~30mins)