<pie_>
i get stuff like do not know how to unpack source archive /nix/var/nix/profiles/per-user/root/channels/nixos/pkgs/applications/virtualization/virtualbox
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjhmr
<pie_>
if i do unpackPhase = ''cp -r $src $out''; it says "cp: cannot stat '/nix/var/nix/profiles/per-user/root/channels/nixos/pkgs/applications/virtualization/virtualbox': No such file or directory"
<pie_>
i kind of get whats going on but im not sure how to fix it
<{^_^}>
[nixpkgs] @andir pushed 5 commits to release-19.03: https://git.io/fjhmi
<pie_>
oh sorry, i had an extra builtins.toPath in there for some reason...without that things are going better
<lopsided98>
judson_: the standard procedure is to submit a PR against master. If it is a bugfix with little chance of breakage it can be backported to 19.03
<simpson>
I don't think that you can, *but* you can configure Wine's Z: drive to point outside.
<freeman[NixOS]>
oh, I found it, nvm
<emily>
(which is a good opportunity to remind that Wine is not a sandbox and any untrusted code you run in it has full permissions over your home directory by default)
myself_ has quit [Remote host closed the connection]
jb_20 has joined #nixos
<jb_20>
How long does it take for a package in nixpkgs on github to be updated in the repo? Specifically the grafana package
<jb_20>
Actual problem: I tried to install grafana but I got a sha mismatch. The version on github right now is correct, though, including on the 19.03 branch, which is my current nixos version
jb_20 has quit [Remote host closed the connection]
<MmeQuignon>
Hi ! Is it possible to create an external nix file to store keys in order to import those keys from another nix file ?
cyraxjoe has joined #nixos
<MmeQuignon>
Actually, I would like to store passwords in a gitignored file and to reference those passwords from versioned files.
cyraxjoe has quit [Client Quit]
cyraxjoe has joined #nixos
<pie_>
growpotk-: growpotkin: attempting to apply patch in overcomplicated manner, its not workng :D https://bpaste.net/show/yo2-
* pie_
runs off to bed
<MmeQuignon>
pie_: 'night !
<pie_>
growpotkin: it executed, but the returned derivation is still the same and virtualbox still gails
cyraxjoe has quit [Quit: I'm out!]
dbmikus has quit [Ping timeout: 258 seconds]
polman has quit [Ping timeout: 245 seconds]
polman has joined #nixos
cyraxjoe has joined #nixos
qubasa_ has quit [Ping timeout: 245 seconds]
qubasa has joined #nixos
<evanjs>
Probably a bad idea, but is there any way to specify a module from a specific channel? Using nixpkgs-unstable and a PR that was recently merged is in nixos-unstable-small
<samueldr>
with great power comes great responsibility
<Miyu-chan>
> <../..
<{^_^}>
error: syntax error, unexpected '<', at (string):263:1
<Miyu-chan>
> <../.>
<{^_^}>
file '../.' was not found in the Nix search path (add it using $NIX_PATH or -I), at (string):263:1
<Miyu-chan>
:D
<samueldr>
it's pretty well whitelisted
<Miyu-chan>
Kinda sad that NIX_PATH will be deprecated by flakes tbh. I've ad-hoc replaced some modules that way. :/
abrar has joined #nixos
felixfoertsch has quit [Ping timeout: 250 seconds]
felixfoertsch has joined #nixos
mkoenig has quit [Remote host closed the connection]
nixy is now known as Guest13501
mkoenig has joined #nixos
nixy has joined #nixos
evanjs| has joined #nixos
Soo_Slow has quit [Remote host closed the connection]
lord| has quit [Ping timeout: 246 seconds]
lord| has joined #nixos
<evanjs>
samueldr: sweeeet so it's just like a normal import, I can just be super specific up to what nix files I want. Though it looks like there's a little extra work if you want to go that route
<samueldr>
evanjs: <something> is more of a special way to refer to a path, so not specifically an import
<samueldr>
you could have src = <mycoolproject>
<samueldr>
and NIX_PATH=mycoolproject=/dev/null
<evanjs>
samueldr: sure sure. I was importing the path and was just thinking of it that way. intrigued to see how flakes will changes things up
freeman42[NixOS] has quit [Ping timeout: 244 seconds]
iqubic` has quit [Remote host closed the connection]
iqubic has quit [Ping timeout: 276 seconds]
iqubic has joined #nixos
<mac10688>
Am I able to just configure mysql from a shell and run it there or do I have to configure my nixos machine to host mysql?
<mac10688>
when i say shell, i mean from a nix file within my project
phreedom_ has joined #nixos
UltraPhil has quit [Quit: Leaving]
phreedom has quit [Ping timeout: 260 seconds]
domogled has joined #nixos
abathur has joined #nixos
<jasom>
clever: it looks like enableDebugging does more than just "dontStrip=true"; I was able to reproduce the issue with debug symbols by just applying dontStrip = true, which is useful for reproducability. Changing optimization flags to gcc makes the problem disappear.
<eyJhb>
Are there any projects that allows for the use of git commits, as channels? So that I can checkout channels at X?
<provessor[m]>
eyjhb: what do u mean
<provessor[m]>
like use packages in a local repository?
<eyJhb>
provessor[m]: like using nixos/nixpkgs stable at a specific commit sha
polman has quit [Ping timeout: 245 seconds]
oida has quit [Ping timeout: 260 seconds]
polman has joined #nixos
<provessor[m]>
I dont know how to do that in your system wide channels apart from hosting the repository yourself at that commit
schjetne has quit [Ping timeout: 244 seconds]
<eyJhb>
Yeah, that was what I was thinking as well.. Just seems like it would be a nice thing to have. But that might just be me
ee1943 has quit [Ping timeout: 245 seconds]
<provessor[m]>
Do you need them as system packages
<provessor[m]>
Would installing as a user be good enough?
<eyJhb>
What do you mean?
<provessor[m]>
I mean system wide packages
ee1943 has joined #nixos
<eyJhb>
Yeah, should be system wide packages
<provessor[m]>
ok yeah I can't help with that
<eyJhb>
Might look at it at some point, as it should be fairly "easy". But then again, I have had problems with Github not playing well
Ariakenom has joined #nixos
<provessor[m]>
It might be enough to download one of the channels (https://nixos.org/channels/) and then find the right git revision, src-url, etc. for the commit you want
<provessor[m]>
then going `nix-channel --add file:///path/to/channel/directory alias`
cfricke has joined #nixos
bakakuna has joined #nixos
oida has joined #nixos
vika_nezrimaya has quit [Ping timeout: 258 seconds]
ixxie has joined #nixos
infinii has joined #nixos
infinee has quit [Ping timeout: 246 seconds]
<provessor[m]>
eyjhb: I'll test that out in a bit and see if it works for u
<provessor[m]>
eyjhb: I suppose that would be very helpful for testing and stability purposes
<eyJhb>
Yeah, e.g. if I want to have two computers in-sync. But I could of course just update them to latest each time.. But, yes
Neo-- has joined #nixos
<provessor[m]>
Yeah just like u would want for something like a server cluster
<eyJhb>
Yeah
<eyJhb>
But I guess, you would always keep those up-to-date. But, probably (__monty__) maybe weeks behind
<eyJhb>
Maybe someone has a insight in why this isn't a thing?
<eyJhb>
How do you handle this at your new workplace srhb , if it is a thing?
<provessor[m]>
I don't know how the current servers are structured but they could make a "channel" that takes a constituent through GET and serves the package tree corresponding to that
<blasternina>
How to make pkgs.writeText ignore tabs? I want to ident my code but cannot have tabs before the line
vyorkin has joined #nixos
hyper_ch2 has quit [Remote host closed the connection]
<alexarice[m]>
blasternina: if nothing else works you could use a file as source
<eyJhb>
Or sed after writing file :p Which is hacky
<alexarice[m]>
though it shouldn't have tabs anyway, how are you using it
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<blasternina>
I mean pkgs.writeText "bla.text" ''
<blasternina>
tab tab my line
<blasternina>
'';
<eyJhb>
blasternina: concrete examples are always nice :D E.g. pastebin
<lassulus>
I think it ignores leading spaces but not leading tabs
<alexarice[m]>
you should probably use '' instead of "
<alexarice[m]>
that's not obvious at all in that font `''` instead of `"`
<blasternina>
yeah ' and ', so two '
<blasternina>
for multi line
mexisme has joined #nixos
o1lo01ol1o has joined #nixos
zupo has joined #nixos
<Mateon1>
How can I check which versions of things changed since a past generation? My setup regressed because of either an amdgpu driver update, xfce or X update.
blasternina has quit [Remote host closed the connection]
<Mateon1>
Now, when one of my monitors turns off, X/xfce automatically switches back to single screen mode, and when I switch the monitor back on, xfce defaults to screen mirroring instead of extending screen area.
<blitzclone_>
hey! I have a PR https://github.com/NixOS/nixpkgs/pull/67975 that seems to build fine by ofborg, but locally get errors fetching one NPM dependency. This dependency was updated recently (18h ago), so there might be some magic to sync the npm packages that hasn't happened yet?
<alexarice[m]>
ofBorg won't evaluate automatically unless you are known user to ofBorg
sb0 has quit [Ping timeout: 258 seconds]
<alexarice[m]>
or if someone tags it in the pr
sb0 has joined #nixos
drakonis has joined #nixos
chloekek has quit [Ping timeout: 245 seconds]
<blitzclone_>
k
<blitzclone_>
let me try
<alexarice[m]>
does anyone know why lots of hydra jobs for unstable are aborting?
hyper_ch2 has joined #nixos
<andi->
some runners are OOM
<andi->
not OOM but out of disk space
<alexarice[m]>
that seems quite bad, does hydra have a way of automatically sorting this out or is it just going to keep failing
MoreTea has joined #nixos
<alexarice[m]>
blitzclone: I don't think that will do anything unless you are on ofBorg's list of users
<blitzclone_>
k. makes sense
<andi->
alexarice[m]: those nodes have auto GC configured. It should work. Yet unknown why it doesn't.
polytux has joined #nixos
ggp0647 has quit [Ping timeout: 244 seconds]
monotux has quit [Ping timeout: 250 seconds]
ggp0647 has joined #nixos
ThatDocsLady has joined #nixos
incognito999 has quit [Ping timeout: 244 seconds]
<Taneb>
What does it mean to get "error: opening lock file '/nix/var/nix/profiles/default.lock': Permission denied" when running nix upgrade-nix (on nix-on-ther-linux)
<sphalerite>
vyorkin: yes, they are, as you can tell from the drv hash being the same. The one with an _ is probably a compatibility alias after renaming it to use only - to fit nixpkgs conventions better.
<sphalerite>
Or something like that.
<vyorkin>
sphalerite: gotcha, thanks
<{^_^}>
[nixpkgs] @Rakesh4G opened pull request #68097 → pythonPackages.labelbox: init at 2.0.1 → https://git.io/fjhnV
o1lo01ol1o has quit [Remote host closed the connection]
joko has joined #nixos
joko has quit [Changing host]
<infinisil>
sphalerite: (well the path was for the one with "-" both times)
<boxdot1>
I need a different mechanism to modify sources instead of patch. Patching just does not apply in my case, I need to do some string replacement. In postUnpack the src cannot be modified anymore. Which phase would be the right one to do it?
<{^_^}>
[nixpkgs] @dylex opened pull request #68099 → graphviz: stable_release_2.40.1 tag disappeared → https://git.io/fjhCu
<symphorien>
boxdot1: by default, the source is copied and made writable in $PWD by the unpack phase
<symphorien>
patch ., not $src
<majtal>
well, tried 19.03 channel, same old. import <nixpkgs> is broken
zupo has joined #nixos
<manveru>
majtal: very strange
majtal has quit [Remote host closed the connection]
veske2 has quit [Quit: This computer has gone to sleep]
<provessor[m]>
<multun "is it because the channel is bui"> fails to build
<provessor[m]>
> <@freenode_multun:matrix.org> is it because the channel is building, or some other reason ?
<provessor[m]>
* fails to build
<{^_^}>
error: syntax error, unexpected '<', at (string):263:1
<provessor[m]>
looks like its qtbase
<provessor[m]>
> <@freenode_multun:matrix.org> is it because the channel is building, or some other reason ?
<provessor[m]>
* fails to build
<{^_^}>
error: syntax error, unexpected '<', at (string):263:1
<provessor[m]>
looks like it because of qtbase
<provessor[m]>
looks like its from qtbase
<nixos-noob>
I'm new to nixos and have been really enjoying it so far, but am going to have to install it on a system which is annoying to connect to the internet. I was wondering if I could create a nixos iso that includes the packages?
<tobiasBora>
but okular, libre office... cannot find the same printers (they do find some), but I can't print with them
domogled has joined #nixos
<tobiasBora>
provessor[m]: both. In firefox & okular, the printer seems different, but the good ones never appear. In chromium, sometimes I need several Ctrl+shift+P before actually seeing the printer, but usually it appears quickly
<provessor[m]>
do they all show correctly in cups?
<etu>
tobiasBora: Well, it will be in next stable branch and whenever nixos-unstable updates manual
<etu>
tobiasBora: But this is how I've set one printer up while I was testing that PR: http://ix.io/3Sf -- not including the enabling of printers and installing drivers part that is needed as well.
<tobiasBora>
provessor[m]: that's a good question. I would say "no", because the printers that appear here have names like "01_02_03_MFP_color_cups" (with underscores and 'cups' at the end), while the one that works with Chromium have names like "0102-03-MFP-LAB-color" (dashed, no 'cups' at the end)
<provessor[m]>
<etu "tobiasBora: Well, it will be in "> It might take a while as nixos-unstable isn't building atm
<nixos-noob>
Ox4A6F: thanks, i'm guessing the answer is yes. I'd better try it before doing it for real
elux has joined #nixos
<elux>
hi there
nixos-noob has quit [Remote host closed the connection]
evanjs| has joined #nixos
<elux>
ive been considering using nixops for my network deployments. just wondering how others' experience has been with nixops and if im someone comfortable with nixos as my daily drvier, if ill feel at home with nixops and things will work as expected?
<tobiasBora>
provessor[m]: the printers with '_' also appear in chromium, but they are greyed with a text like "pause: all jobs are rejected"
<tobiasBora>
provessor[m]: if it can help, I think that "http://cups.mylab.fr:631/printers/" list the good printers. I've no idea how chromium manage to get them and not okular & cups.
<provessor[m]>
tobiasBora: the problem might be that the printers arent configured properly system wide
<__monty__>
,locate bin write-good
<{^_^}>
Couldn't find in any packages
<provessor[m]>
cause chromium does its own thing instead of using a lot of services and cups might be one of them
romildo has joined #nixos
romildo has left #nixos [#nixos]
domogled has quit [Remote host closed the connection]
<lassulus>
is nixos-unstable borken? i get errors about vulkan
<provessor[m]>
I can't come up with any suggestions tho, might have to wait for someone else who knows about printing to come on
<tobiasBora>
provessor[m]: note that from the cups.mylab.fr website, I tried to manually add a printer using the lpd://0102-03-MFP-LAB.common.mylab.fr/raw
<tobiasBora>
but when I try to print, I get an error Unable to locate printer cups.local
<misuzu>
lassulus: seems like that, i had to remove mpv as it wants vulkan-loader, but it is not building
<tobiasBora>
provessor[m]: I think the printers with '_' are badly configured, but the ones with '-' are configured correctly on the network... Now I don't know why they appear only in chromium... But I have no idea how cups "propagate" the printers...
<elux>
clever: nice and simple :)
<tobiasBora>
hum... Something new here!
<clever>
elux: yep
<provessor[m]>
how did you add the printers to begin with
hyper_ch2 has quit [Remote host closed the connection]
<elux>
although, i do enjoy using docker because it helps with our CI process, and log aggregation etc. i assume nixops would be friendly to work with docker containers too?
<elux>
might lose some of its benefits tho.. but perhaps for network+machine provisioning nixops can work well for this, deploying containers tho, hopefully can help too..
<tobiasBora>
so I tried to add the printer directly from cups, and he can find some printers that work with dnssd:// address. I also just open the port 631 and now I can print with them (no advanced configuration available however, I'll try to change the driver)... Let's try again
<tobiasBora>
(for now driver is PageWide PageWide ... IPP everywhere)
<provessor[m]>
tobiasBora: good point, I didn't think about ports
<Ariakenom>
I was looking into cpuFreqGovernor in hardware_config.nix and there was a lib.mkDefault call and I don't know what it means
<clever>
Ariakenom: that means you can set it again in configuration.nix and it wont conflict
<clever>
Ariakenom: because you should generally avoid editing hardware-configuration.nix, since nixos-generate-config overwrites it
philr has joined #nixos
domogled has joined #nixos
ixxie has joined #nixos
zupo has joined #nixos
<tobiasBora>
provessor[m]: firefox seems to detect sometimes the printer now, with all the options. With a different driver, I can add with cups the printer and it seems to work, but I can't set the stapler option while I can on firefox and chromium... and the driver seems to be the same as the one at cups.lip6.fr...
npmccallum has joined #nixos
<tobiasBora>
and I'm not sure if I can print because I opened the port, or because I restarted cups.
<tobiasBora>
Because I noticed a few months ago that I needed to restart cups to make it work
veske has joined #nixos
<Ariakenom>
clever: Thanks.
<elux>
mdash: nice morph looks cool too
<provessor[m]>
tobiasBora: Is this just over a local network
veske has quit [Client Quit]
yorick has joined #nixos
<Ariakenom>
I have a hard time figuring out what I should set it too though
dasj19 has quit [Read error: Connection reset by peer]
dasj19 has joined #nixos
<tobiasBora>
provessor[m]: I don't know... To see the working printers I need to go on cups.mylab.fr:631, so I'm thinking that some external network may be involved, but don't know how chromium is aware of that
<sm[m]>
G’day all. I think most Haskell packages don’t have their own .nix file in nixpkgs, and are not updated individually; rather nixpkgs automatically includes them when they appear in stackage nightly. Is that right ?
<provessor[m]>
tobiasBora: it could still be local
<mdash>
elux: i'm liking how krops does secrets management
<provessor[m]>
the mylab.fr makes my think it probably is local
<mdash>
Speaking of secrets management, what are people doing these days to keep secrets out of the store? some nixos services don't seem to have a good way to do that, should those just go in containers? would that even help?
romildo has joined #nixos
<__monty__>
sm[m]: Stackage+head.hackage I think.
hyper_ch2 has joined #nixos
romildo has left #nixos [#nixos]
<tobiasBora>
etu: in the link you sent me, how do you determine the field "postscript-lexmark/Lexmark-CS510_Series-Postscript-Lexmark.ppd"? In my case the driver I need is HP PageWide Color E77650-E77660 Postscript, but I don't know how to translate to this .ppd string, especially because two drivers have the exact same name (and I need the "second" one)
dansho has quit [Ping timeout: 244 seconds]
<sm[m]>
Ah, thx
jmeredith has joined #nixos
zupo has quit [Ping timeout: 244 seconds]
<tobiasBora>
provessor[m]: ok, then I don't know... I just now that the printer is reachable with dnssd://01-02%2F03%20MFP%20LAB%20(color)%20%40%20cups._ipp._tcp.local/cups?uuid=cdf1ba94-1039-38d3-4e11-111111111111
<sm[m]>
Also, where can I find that shell script that adds .nix-profile/bin to $PATH ? I can never find it again
dansho has joined #nixos
<provessor[m]>
<tobiasBora "provessor: ok, then I don't know"> If it works thats all good, I was just wondering if u knew for in the future if a problem like this happens again
<tobiasBora>
provessor[m]: I'm still not 100% sure to understand everything that happens ^^ First, I *need* to use cups to install the printer, KDE's interface for new printers does not work. Any idea why?
peel_ has joined #nixos
elux has quit [Quit: leaving]
<etu>
tobiasBora: I think I looked in the driver package and guessed from other examples
<tobiasBora>
etu: and do you know if the name is relative to some path, or it's an absolute name?
<etu>
tobiasBora: It's relative
<etu>
tobiasBora: If I build that drivers package postscript-lexmark
peel has joined #nixos
<etu>
tobiasBora: The path should be the the directories below result/share/cups/model/
<evanjs>
Is there a pratical way to query hydra for builds and see how a package is making its way across channels?
<multun>
^
<provessor[m]>
<tobiasBora "provessor: I'm still not 100% su"> on https://community.kde.org/Printing#Current_information it says KDE uses the Qt printing infrastructure which might just use cups as a backend but I'm not sure
<provessor[m]>
their printer management page hasn't been made yet so maybe it's not implemented correctly atm
__monty__ has quit [Read error: Connection reset by peer]
<{^_^}>
[nix] @edolstra pushed 0 commits to 2.3-maintenance: https://git.io/fjhl7
__monty__ has joined #nixos
jtojnar has joined #nixos
<__monty__>
clever: I have tuntaposx installed. I think that's what you were referring to with openvpn tun driver?
gentauro has quit [Read error: Connection reset by peer]
gentauro has joined #nixos
mexisme has joined #nixos
<__monty__>
clever: Oh, wait a minute. The tun0 error was probably due to permissions. /facepalm
<clever>
__monty__: oh, lol
<clever>
__monty__: the `-u foo` assumes you run it as root, and it will drop-root on its own after opening things
<__monty__>
clever: Hmm, with the -u flag I get these ip related permission errors: http://ix.io/1Uf2 Any way around those or do I need to run it as root?
mexisme has quit [Ping timeout: 245 seconds]
<clever>
__monty__: ah, maybe i never tested -u on mac
<clever>
just let it run as root
Ilya_G has joined #nixos
<__monty__>
clever: Hmm, I have another box that claims to be connected udp but I can't ping/ssh to the "friend."
<__monty__>
I added that machine to toxvpn on the mac as a friend to test but that's not working so I'm troubleshooting that one first.
<Ilya_G>
Complete NixOS Noob here and could really use sombody's help understanding the ecosystem.
<__monty__>
Might double NAT interfere with toxvpn's working?
<__monty__>
,ask Ilya_G
<{^_^}>
Ilya_G: Just ask your question. It's the best way to know if anybody can help. Or look at other ways of getting in touch with the Nix community here: https://nixos.wiki/wiki/Get_In_Touch
<Ilya_G>
I am working from MacOS and I am trying to build a NixOS virtual machine image for VMWare
<Ilya_G>
I got nixos/nix docker running
<Ilya_G>
I am able to build an image there from a sample I found on a blog post out there.
<Ilya_G>
I am literally confused about the workflow of this ordeal.
<clever>
`This is a demo appliance for VirtualBox (in OVA format) that has X11 and Plasma 5 enabled, as well as the VirtualBox guest additions.`
<Ilya_G>
__monty__ I am Noob in more than one way here :D Thanks for clarifying.
<Ilya_G>
clever I'll start from there. I need ultimately to produce headless image with with just ssh in it as I will have no access to screen nor do I need it.
<Ariakenom>
will auto gc at some points run if the system isn't running during nix.gc.dates?
<clever>
Ariakenom: nope
dasj19 has quit [Ping timeout: 252 seconds]
MmeQuignon has joined #nixos
hyper_ch2 has joined #nixos
<Ariakenom>
oh wait. It's set to "weekly". That ought to work.
zfnmxt has quit [Quit: Bye!]
zfnmxt has joined #nixos
<{^_^}>
[nixpkgs] @jtojnar opened pull request #68105 → meson: fix darwin patch → https://git.io/fjh8n
Makaveli7 has quit [Quit: WeeChat 2.5]
hyper_ch2 has quit [Remote host closed the connection]
<tobiasBora>
etu: I found command "lpinfo -m" which is quicker to find the list of drivers, with the url to give to cups
<tobiasBora>
and the advantage of this command is also that it prints the description
<tobiasBora>
so you can actually grep it
<tobiasBora>
for example, I realized that the description "HP PageWide Color E77650-E77660 Postscript" was bound to both "HP/hp-pagewide_color_765-ps.ppd.gz" and "HP/hp-pagewide_color_e77650-e77660-ps.ppd.gz"
waleee-cl has joined #nixos
Ilya_G has quit [Remote host closed the connection]
<__monty__>
clever: Ok, what am I looking for in tcpdump output? I see the outgoing requests I think but nothing comes back. Can't check the other side because that's remote. Or did you mean between the two local computers?
zfnmxt has quit [Quit: Bye!]
zfnmxt has joined #nixos
MmeQuignon has quit [Ping timeout: 245 seconds]
<waleee-cl>
what is the status for the vulkanloader assertion-error on nixos-unstable? Is it still blocked by failing tests?
<clever>
__monty__: run the tcpdump on the tun interface, of both machines, local and remote, does something going in one end, come out the other?
<clever>
waleee-cl: i have seen nixos-unstable-small update a few times since then
<srhb>
waleee-cl: iirc it's fixed in master
<waleee-cl>
clever: yeah, I saw that, but apparently vulkanloader isn't in the small-channel
veske has quit [Quit: This computer has gone to sleep]
<waleee-cl>
srhb: ok, I still gets the same assertion error though after updating the channels and rebuilding
<srhb>
waleee-cl: The channel didn't bump yet, because th eval isn't complete
zfnmxt has quit [Quit: Bye!]
<srhb>
waleee-cl: But that package in particular is good, whether or not it fails to update for other reasons once it's done.
<blackriversoftwa>
Is there a good way to mark a derivation as needing lots of memory, and keep it from building in parallel with other derivations marked that way?
<srhb>
(Assuming that was the issue in question...)
nakkle has joined #nixos
zfnmxt has joined #nixos
<blackriversoftwa>
(we have a cross-compile that uses qemu binary emulation to run some Template Haskell, and because qemu's binary emulation still allocates a big fixed chunk of memory, our build server's kernel keeps randomly killing these builds if they run in parallel and take up all of the build server memory)
<__monty__>
clever: Hmm, weird. So two local hosts mac and lin. lin is tethered using ethernet to mac, mac is connected to internet using wifi. Added lin as a toxvpn friend on mac. lin can ping/ssh to mac but not the other way around. However, toxvpn-remote doesn't notify about a friend request and I haven't whitelisted/added mac manually.
<waleee-cl>
srhb: ah, I might have been a tad imprecise. "Is the channel still blocked from advancing to that commit by failing tests (... for other services/packages)"
<srhb>
waleee-cl: See the two leftmost columns (the newest)
fendor has quit [Ping timeout: 258 seconds]
<srhb>
waleee-cl: None of them have any failures. If they all turn green, once the eval completes (all packages have been attempted to be built on top of those tests) -- it will bump
<srhb>
waleee-cl: If they turn red, then "yes"
<nakkle>
Hey there! I have a nixos configuration, where I extended the argument set by `hardware-configuration ? ./hardware-configuration.nix`, and put this into imports like this: `imports = [ hardware-configuration ];`. This causes an infinite recursion error when I do nixos-rebuild. If I do `import = [ ./hardware-configuration.nix ]`, it works fine. What exactly is the difference between putting it in there as a variable vs as the raw path?
<srhb>
waleee-cl: Or, to put it differently, we don't know yet. :-)
<srhb>
It's always a moving target.
<waleee-cl>
srhb: thanks. I knew about hydra, but I hadn't kept track of which commit, jobset &c the vulkan-loader fix was in
<srhb>
waleee-cl: Sure thing.
byteflam1 has joined #nixos
<srhb>
waleee-cl: fwiw I opened the leftmost one, checked its corresponding eval, looked up vulkan-headers, and it was green. From that I can tell that iff the tested job completes succesfully, the fix will be in the channel bump..
<srhb>
er, loader, not headers..
<{^_^}>
[nix-mode] @matthewbauer merged pull request #91 → Don't freeze full executable paths at load time → https://git.io/fjh3j
<{^_^}>
[nix-mode] @matthewbauer pushed 2 commits to master: https://git.io/fjh8b
FRidh has joined #nixos
<{^_^}>
[nix-mode] @matthewbauer pushed 3 commits to master: https://git.io/fjh8N
<__monty__>
clever: There's nothing in tcpdump of the tun on either side when I try mac and lin. Maybe tun is skipped for ethernet?
<o1lo01ol1o>
I keep getting loops when trying to download derivations from a cache: file 'nar/klpf8njb8gpgkzs260pr5rbs63vcmcsx.nar' does not exist in binary cache 'foo'. And then nix proceeeds to copy the path from `foo` only to find some .nar doesn't exist there. This will happen for 3-4 derivations for hours, does anyone know what might be going on?
<boxdot1>
I am confused by one thing: my default.nix has as input a hook script. Whenever, I change it the package is not rebuilt. I thought the content of the script is part of the derivation.
freeman42[NixOS] has joined #nixos
<boxdot1>
I speak about cargoUpdateHook in fetchcargo.nix
DariusTheMede has joined #nixos
<clever>
__monty__: sounds like your using the wrong ip
<clever>
__monty__: does toxvpn show the 2 peers as being connected, and having non-conflicting IP's?
domogled has quit [Ping timeout: 245 seconds]
<__monty__>
clever: No, like I said, I added lin on mac, but not the other way around.
<evanjs>
Is there an RFC for, or is it even _possible_ to determine closure since before realization, currently? Not certain those terms are totally correct *shrugs*
<evanjs>
Motivation for me at least would be low disk space + telling me I can't do a full system rebuild beforehand >.>
<clever>
__monty__: you cant exchange packets until toxvpn says the other peer is online
<Taneb>
evanjs: I don't believe that that's possible or even in general reasonable
<clever>
__monty__: what ip's are you using, and what is your lan?
emptyflask has joined #nixos
<Taneb>
evanjs: it might be possible if it's fetching everything
<evanjs>
Taneb: Hrm. What about expression-level checks like "you need at least ___ of free space"?
<__monty__>
clever: Yes, that's how I assumed it'd work. Hence why I'm confused. mac 192.168.0.177, lin 192.168.2.4, mac tox 10.0.0.4, lin tox 10.0.0.10
<Taneb>
evanjs: that's an option, but that sort of checks are hard to keep up to date, and if they're above or if they're below then they can cause frustration
<__monty__>
clever: mac can't ping 10.0.0.10, lin can ping/ssh to 10.0.0.4. (I have since added mac on lin but neither's showing connected. Wanted to see whether adding someone one-sided was enough to allow incoming connections.)
<{^_^}>
[nixops-aws] @AmineChikhaoui pushed 0 commits to refs/tags/v1.0.0: https://git.io/fjh4Z
<__monty__>
clever: mac and lin are connected with a cable, mac is connect to my wifi.
<clever>
__monty__: that sounds like its just the nixos firewall, can mac ping 192.168.2.4?
<evanjs>
Taneb: I can see that. Just wondering how I can prevent myself from rebuilding for a while until I get stopped because of low disk space, clear up stuff, nix-collect-garbage -d, repeat...
domogled has joined #nixos
<clever>
__monty__: if ssh works one way, then the vpn is fully working, and its just firewalls
<{^_^}>
[nixops-hetzner] @AmineChikhaoui pushed 0 commits to refs/tags/v1.0.0: https://git.io/fjh4C
<__monty__>
clever: Ok. But why could I ssh to mac *before* adding/whitelisting anything on lin?
<clever>
__monty__: linux rarely blocks outgoing traffic
vika_nezrimaya has joined #nixos
schjetne has quit [Ping timeout: 245 seconds]
byteflam1 has quit [Ping timeout: 244 seconds]
<__monty__>
clever: Doesn't toxvpn use a random port? Can I configure the firewall to allow incoming traffic to toxvpn?
<__monty__>
Or would I allow all incoming traffic on the tun interface?
<clever>
__monty__: toxvpn uses udp, and the firewall lets replies to udp come back in, it treats it like an outgoing connection
<{^_^}>
[nixpkgs] @jtojnar pushed 2 commits to staging-next: https://git.io/fjh4l
<clever>
__monty__: and most systems dont block outgoing traffic, or the replies to that
byteflam1 has joined #nixos
<nakkle>
Hey there! I have a nixos configuration, where I extended the argument set by `hardware-configuration ? ./hardware-configuration.nix`, and put this into imports like this: `imports = [ hardware-configuration ];`. This causes an infinite recursion error when I do nixos-rebuild. If I do `import = [ ./hardware-configuration.nix ]`, it works fine. What exactly is the difference here between putting it in there as a variable vs as the raw path
<nakkle>
?
domogled has quit [Remote host closed the connection]
<clever>
nakkle: the imports value must not depend on the args the module receives, because the args you receive can come from other modules in imports
<__monty__>
clever: Then why doesn't toxvpn list the friend as connected? And why can't I ping/ssh to lin from mac? I thought you identified the firewall as the problem?
Fluffy[m] has left #nixos ["User left"]
iyzsong has quit [Ping timeout: 264 seconds]
<clever>
__monty__: the firewall can block pings over the vpn, after the vpn itself is working
<nakkle>
clever: I don't think I understand. If I put this config into a node of a nixos test, and overload the hardware-configuration argument, everything is fine. If I do it on a real nixos machine, the hardware-configuration should be hardcoded to the hardware-configuration.nix, but it throws the recursion error. Is there a difference between evaluating configs for nixos tests and real machines?
<__monty__>
clever: Wouldn't the toxvpn's list eachother if the scenario was as you think it is though?
<clever>
nakkle: depends on if its a module level arg, or a function outside the module
<clever>
__monty__: yeah
<nakkle>
clever: I still don't understand. what is a module level arg? the `hardware-configuration` argument I put into the argument set?
freeman42[NixOS] has quit [Ping timeout: 250 seconds]
<DariusTheMede>
nix-shell -I nixpkgs=~/nixpkgs -p "haskellPackages.ghcWithPackages (pkgs: [pkgs.accelerate pkgs.accelerate-llvm pkgs.accelerate-llvm-native pkgs.accelerate-fft])" --arg config '{ allowUnfree = true; }' still gives me "unfree license (‘unfree’), refusing to evaluate"
<__monty__>
clever: That's not happening. I don't know why, maybe because the tethered host isn't reachable from the internet? (I'm assuming mac goes out to tox to find lin, maybe that return path doesn't work?) But I'm still perplexed lin can ssh to mac without having added/whitelisted mac.
<clever>
__monty__: if there are nat troubles, toxvpn will use a tcp relay as a fallback
<DariusTheMede>
clever: :thumbsup:
<__monty__>
clever: Any other way to explain lin showing as connected to lin2 but not being able to ping/ssh and lin showing as *not* connected to mac but being able to ping/ssh?
byteflam1 has joined #nixos
<clever>
__monty__: can you make a gist showing the output of `list` and `status` from each machine, and also the ip's of everything, and what directions work where?
jmeredith has quit [Quit: Connection closed for inactivity]
<infinisil>
nakkle: Not sure why you'd want to put hardware-configuration as a module argument, nixops doesn't provide you with it
<infinisil>
What are you really trying to do?
dsx has quit [Quit: dsx]
Gigabitten has quit [Ping timeout: 258 seconds]
freeman42[NixOS] has joined #nixos
peel_ has quit [Quit: Connection closed for inactivity]
emptyfl4sk has joined #nixos
<nakkle>
infinisil: the idea is to have a configuration for a server, and write a nixos qemu test which contains a node that's defined by that exact configuration. but within a nixos qemu test, I don't have a hardware-configuration.nix file, but might want to overload it with, e.g., virtualisation.memory = someValue.
emptyfl41 has joined #nixos
Chiliparrot has joined #nixos
vika_nezrimaya has quit [Ping timeout: 245 seconds]
philr has quit [Ping timeout: 250 seconds]
emptyfl4sk has quit [Ping timeout: 258 seconds]
DariusTh` has joined #nixos
dsx has joined #nixos
DariusTheMede has quit [Ping timeout: 244 seconds]
jmeredith has joined #nixos
<judson_>
Does anyone have githooks for the Nixpkgs commit format?
<{^_^}>
#55786 (by cleverca22, 28 weeks ago, closed): improve perl shebang lines by switching to `use lib ...;`
<__monty__>
Ran into a patched perl shebang that was really long earlier today. Why doesn't nix respect the fact that shebangs can only have one command and one argument?
beyuu has quit [Remote host closed the connection]
Neo-- has joined #nixos
<{^_^}>
[nixpkgs] @Infinisil merged pull request #68015 → nixos/zookeeper: recursively set permissions and ownership on dataDir → https://git.io/fjhvd
<Fare>
Hi. I have a private git repo. What's the recommended way to deal with it? Using the URL with fetchgit won't work because it's run from a sandbox without my credentials.
evanjs^ has joined #nixos
<clever>
Fare: builtins.fetchGit
<ivan>
srhb: not to worry, I git revert -m all the time
<Fare>
Also, for casual builds, I'd like to build from my local repo, I suppose.
<ivan>
yorick: ah
<ivan>
Fare: I use a cgit running on a wireguard interface
evanjs| has quit [Ping timeout: 264 seconds]
evanjs^ has quit [Read error: Connection reset by peer]
evanjs| has joined #nixos
<{^_^}>
[nixpkgs] @srhb opened pull request #68114 → nixos/manual: Fix build after broken Ceph release note → https://git.io/fjhdU
schjetne has quit [Ping timeout: 246 seconds]
Fare has quit [Ping timeout: 264 seconds]
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos
<ashkitten>
how can i make my system reboot before running a systemd service on a timer?
<ashkitten>
zpool scrubs always lock up my system if i don't reboot first
<ivan>
turn the timer off and scrub your system after boot instead?
<srhb>
ashkitten: touch a file, make a boot thingy that has PathExists, then clears it after scrub?
<srhb>
But.. That doesn't sound nice :P
<ivan>
alternatively don't use filesystems that lock up your machine :/
<ashkitten>
i'm confident it'll be fixed soon enough
dbmikus has joined #nixos
<srhb>
ashkitten: Oh? I also have the issue and just went with the zen approach of "even good filesystems are bad" -- any reason to suspect it's fixed?
<{^_^}>
zfsonlinux/zfs#7553 (by ppwaskie, 1 year ago, open): After a week of running array, issuing zpool scrub causes system hang
<ashkitten>
and i didn't say soon, i said "soon enough"
byteflam1 has joined #nixos
<ashkitten>
which to me means something along the lines of "people know about it and it's probably not gonna just get left sitting forever"
ThatDocsLady_ has quit [Remote host closed the connection]
ThatDocsLady_ has joined #nixos
<srhb>
ashkitten: I was probably projecting too much hope into that, sorry :)
<ashkitten>
hah
<ashkitten>
all good
<srhb>
Also english is hard, I didn't think of the distinction. :P Still, that sounds more like what I'm seeing.
<ashkitten>
ughhhh cargoSha256 mismatches
<srhb>
I've actually been meaning to get some useful logs if possible, but once it happens I'm usually too late to get any useful binaries for debugging.
<srhb>
Maybe I should just load some into a ramfs...
thc202 has quit [Ping timeout: 250 seconds]
<ashkitten>
that's probably a good idea tbh
<ashkitten>
your filesystem locks up, yeah? not anything else?
<srhb>
Well, everything becomes unresponsive after a while, but I sort of expected that to be caused by the locked file system, as in after a while things get janky after being unable to write to logs or whatnot.
<srhb>
But I'm honestly not sure.
<srhb>
The first sign will usually be me trying to execute something and it just hangs.
<ashkitten>
yep
<ashkitten>
can't even open the terminal ;-;
sigmundv has quit [Ping timeout: 244 seconds]
<srhb>
Right.
<ashkitten>
and for me i can't switch ttys because physlock can't be started to unlock tty switching
<srhb>
But it's probably even easier to build a simple drv with all the debug tools...
<ashkitten>
yeah i just don't want the rest of my system to lock up while i do it
Fare has joined #nixos
vyorkin has quit [Ping timeout: 245 seconds]
<ashkitten>
so if i could basically switch into a system from the tmpfs
ThatDocsLady_ has quit [Remote host closed the connection]
beyuu has quit [Ping timeout: 260 seconds]
chr15 has joined #nixos
beyuu has joined #nixos
vmandela has quit [Quit: Leaving]
emptyfl41 has quit [Ping timeout: 245 seconds]
polman has quit [Ping timeout: 245 seconds]
polman has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<eacameron>
Is there anything that will let you create a sort of "--help" output for a `default.nix`? I'd like something that lets me do `nix-build -A help` and it `error`s with a generated message for possible targets, generated from the derivation itself.
MmeQuignon has quit [Ping timeout: 245 seconds]
<eacameron>
If not...it's time we write it.
<srhb>
eacameron: Not that I know of, and it's the first time I hear of the idea. It's a really good idea though. :)
<eacameron>
srhb: I'll submit something :D
<srhb>
\o/
<infinisil>
eacameron: tab completion?
<eacameron>
infinisil: yeah that's something at least...
<eacameron>
But I'm imagining a little wrapper DSL that lets you add help descriptions etc. and is also recursive
<infinisil>
You could also make a -A help yourself
<eacameron>
Yeah I plan too..but after I add this 😂
Fare has quit [Ping timeout: 276 seconds]
zxcvbnm7358 has joined #nixos
byteflam1 has joined #nixos
<eacameron>
something like `{pkgs, ...}: let ... in pkgs.withHelp (help: { attr1 = help attr1 "The first thing you might want"; attr2 = { sub1 = help sub1 "The second thing you might want."; }; }`
__monty__ has quit [Read error: Connection reset by peer]
<eacameron>
I wonder if we could reuse the module system somehow...
<{^_^}>
[nixpkgs] @livnev opened pull request #68115 → perlPackages.GlibObjectIntrospection: needs cairo to build test libs → https://git.io/fjhd9
ddellacosta has quit [Ping timeout: 245 seconds]
pro-src has joined #nixos
<boxdot>
Is it a known issue when running `cargo update`: warning: spurious network error (2 tries remaining): [6] Couldn't resolve host name; class=Net (12)
<{^_^}>
[nixos-homepage] @edolstra pushed 2 commits to master: https://git.io/fjhFw
zeta_0 has joined #nixos
emptyfl41 has joined #nixos
<zeta_0>
last time i tried to change home-managers path it threw error when i did a rebuild, i want to be able to access home.nix quickly from from my `/home/user/` quickly with emacs, so what is the exact path that i need to specify?
acarrico has quit [Ping timeout: 258 seconds]
elibrokeit has quit [Quit: A random quit message]
hpfr has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjhFD
dansho has quit [Remote host closed the connection]
dansho has joined #nixos
Myhlamaeus has joined #nixos
Saa77Mo has joined #nixos
qubasa has joined #nixos
sigmundv has joined #nixos
polman has quit [Ping timeout: 245 seconds]
byteflam1 has quit [Ping timeout: 246 seconds]
trubydoor has joined #nixos
<trubydoor>
I'm having an issue today where it seems the version of vulkan-loader isn't the same as vulkan-headers, so I can't do a nixos-rebuild. Anyone else getting this or is something wrong with my setup?
polman has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<trubydoor>
in master the versions seem to be the same though.. how often is nixpkgs-unstable updated from master?
<clever>
sqlite> select path, narsize/1024/1024 as narmb, datetime(registrationTime,'unixepoch') from ValidPaths where path like "%BindingAttributes%" order by registrationTime;
emptyfl41 has joined #nixos
<ingenieroariel>
Hello everyone, I started running nixos in hardened mode yesterday and now I am not able to start sway from the commandline. I am not sure exactly which of the lines in hardened mode does it or how to make an exception
<ingenieroariel>
Here is the error: Session 1 isnt a graphical session (type: tty) direct-ipc.c:35 do not have CAP_SYS_ADMIN; cannot become DRM master
<clever>
boxdot: that will list everything in /nix/store (run it after the warning is done) and sort it by when it was added, and filter oh wait, wrong sql
<clever>
sqlite> select path, narsize/1024/1024 as narmb, datetime(registrationTime,'unixepoch') from ValidPaths where narsize > 256 * 1024 * 1024 order by registrationTime;
<clever>
boxdot: and filter it to greater then 256mb
<trubydoor>
clever: I see. Well, any way for me to fix this temporarily until nixpkgs gets updated? I checked my local nixpkgs and the versions definitely don't match so vulkan_loader can't build
<clever>
Trubydoor: remove the thing that depends on vulkan from your systemPackages
<trubydoor>
clever: I think that'd be my graphics drivers, which I'd rather not remove :)
<boxdot>
clever: very nice. Thanks a lot.
<clever>
Trubydoor: have you tried the nixos-unstable-small channel?
<trubydoor>
I guess there's no (easy?) way for me to just say for now that I specifically want to get vulkan-headers from master
<clever>
Trubydoor: you would make an overlay to change vulkan and vulkan-headers
fendor has joined #nixos
o1lo01ol1o has quit [Ping timeout: 268 seconds]
<trubydoor>
I guess I don't understand enough of this... I tried to do let master = import (fetchTarball...) in ... and then use master.vulkan-headers but I guess that's not the right way to do things
<clever>
Trubydoor: you would need to do vulkan-headers = master.vulkan-headers; in an overlay
ingenieroariel has quit [Remote host closed the connection]
growpotk- has joined #nixos
_zincy has quit [Ping timeout: 252 seconds]
byteflam1 has quit [Ping timeout: 244 seconds]
casaca has quit [Ping timeout: 245 seconds]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
emptyfl41 has quit [Ping timeout: 246 seconds]
Ilya_G has joined #nixos
Ilya_G is now known as Ilya_G_NixOS_Noo
Ilya_G_NixOS_Noo is now known as Ilya_G_NixOSNoob
<worldofpeace>
marek: I'd open an issue with details of what's broken and mention that it's blocking that channel. And mention the qt maintainers and the darwin team
<worldofpeace>
My personal knowledge of both those things combined is pretty much 0%. But that's pretty much how to deal with that situation once you notice it.
<worldofpeace>
And it does actually help a lot to encourage things to keep moving.
Shell has joined #nixos
<evanjs>
Anything else I might need to do for #67115?
Soo_Slow has quit [Remote host closed the connection]
boxdot has quit [Quit: Leaving.]
boxdot has joined #nixos
boxdot has left #nixos [#nixos]
Ilya_G_NixOSNoob has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Ma27 opened pull request #68121 → nixos-option: don't break if `builtins.trace` is used in `<nixos-config>` → https://git.io/fjhAV
byteflam1 has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @worldofpeace closed pull request #54415 → nixos/gnome3: split into common, flashback and 3 → https://git.io/fhzEZ
byteflam1 has joined #nixos
zeta_0 has quit [Quit: rcirc on GNU Emacs 26.1]
joepie91___ has joined #nixos
joepie91___ has joined #nixos
joepie91___ is now known as not-joepie91
<fresheyeball>
is there a way to determine why nix is building something?
<evanjs>
worldofpeace: I'm not seeing any examples of adding libPath to buildInputs as you mentioned. Is this actually something that's done to other binary packages?
<fresheyeball>
like I have a project, I do a nix-build, I don't expect it to build, but it does
<fresheyeball>
how can I determine why
<ldlework>
pretty much the opposite problem i usually have
trubydoor has quit [Remote host closed the connection]
wfranzini has quit [Remote host closed the connection]
<Fare>
what is the right way to use nix-shell with a git checkout that I modify?
<fresheyeball>
Fare: what do you mean?
erasmas has quit [Quit: leaving]
<fresheyeball>
you can just use nix-shell in your project after editing code
<Fare>
well, first off, it's a private git repo, so using fetchGit fails by lack of credentials
<Fare>
then, I installed a lot of npm packages and that would take time to redo every time.
<evanjs>
Fare: fetchGit actually does work with private repos if you have stuff in your ssh config IIRC
<evanjs>
Fare: no submodules though
<evanjs>
gah it wasn't ssh config though was it
<fresheyeball>
Fare: what are you trying to do more broadly
<fresheyeball>
?
<Fare>
evanjs, I do have a ssh config, but fetchgit didn't work with it; I'll retry with fetchGit.
<Fare>
fresheyeball, I want to have my usual build through Nix, so my deployments are deterministic and I can distribute code more easily.
domogled has quit [Ping timeout: 244 seconds]
<Fare>
But I also want my usual build to be fast and incremental.
<worldofpeace>
evanjs: the glib setup hook populates GSETTINGS_SCHEMA_PATH env var which wrapGAppsHook uses to wrap the binary. So these libs need to be in buildInputs for this to happen, and any electron apps that aren't doing this in nixpkgs are likely broken.
<Fare>
I use various languages: Haskell, OCaml, Gerbil Scheme, PLT Racket, JavaScript. Formerly, Common Lisp.
shibboleth has quit [Quit: shibboleth]
<selfsymmetric-mu>
fresheyeball: I would also like to see scoped logs in nix builds, so that a particular line of output would be prefixed by `building gcc-<version>` or whatever.
<selfsymmetric-mu>
Today I can only guess based on the output.
szicari has quit [Ping timeout: 264 seconds]
<fresheyeball>
Fare: ok great
<evanjs>
worldofpeace: Yeah I saw glib exclusively in a few places and that seemed to fix the file picker problem. Not sure what else needs to go in there though
<fresheyeball>
what are you currently doing to serve this goal?
<fresheyeball>
Fare: are you looking for just general project structure advice?
<worldofpeace>
evanjs: By virtue of them being in buildInputs should be enough. You could also you this as an opportunity to use `autoPatchelfHook` which automatically patchelfs things.
byteflam1 has quit [Ping timeout: 268 seconds]
growpotkin has quit [Read error: Connection reset by peer]
<Fare>
fresheyeball, yes, general advice, and in particular how to integrate nix and git for everyday workflow
<worldofpeace>
and autoPatchelfHook needs the libs in `buildInputs`. and technically they should be in buildInputs anyway
growpotkin has joined #nixos
<fresheyeball>
Fare: ok!
<fresheyeball>
so each derivation in your build can be built independently
<evanjs>
worldofpeace: welp. autopatchelf is complaining about a bunch of stuff. Assuming I can use that as a guide of what needs to go into inputs?
<fresheyeball>
so breaking your project into sensible parts will make your builds more incremental
<Fare>
"problem" being that I may want to "open" or "close" some sub-projects, then, and override the defaults with local checkouts being worked on
polman has quit [Ping timeout: 245 seconds]
<fresheyeball>
Fare: are your sub projects seperate repos?
polman has joined #nixos
szicari has joined #nixos
<worldofpeace>
evanjs: autoPatchelfHook fails if there's a library that something needs that isn't in the environment. And yes you can use that as a guide as to what should be added to buildInputs
__monty__ has quit [Quit: leaving]
justanotheruser has joined #nixos
trubydoor has joined #nixos
trubydoor has quit [Client Quit]
<evanjs>
worldofpeace: and it's better to add only what needs to go into buildInputs rather than just buildInputs = [] ++ libPath, right?
<Fare>
fresheyeball, yes, I currently have two main repositories
<Fare>
(plus, sometimes, patches to upstream libraries)
<fresheyeball>
Fare: ok, and are you using npm to install packages in addition to nix?
<Fare>
yes, and stack for haskell, and opam for ocaml, etc.
<fresheyeball>
ok that is your first problem
<worldofpeace>
`libPath` should be uneeeded. Everything can go in buildInputs (of course not the things that are native though)
<worldofpeace>
^ evanjs
<fresheyeball>
you can't have determinism with nix and do that
<fresheyeball>
you need to use nix to get those packages
<Fare>
so how do I migrate my stack/opam/npm/raco/quicklisp packages to nix?
<fresheyeball>
and I would make each part it's on seperate derivation
polman has quit [Ping timeout: 245 seconds]
<Fare>
yes, for a start
<fresheyeball>
Fare: I don't know all of those
polman has joined #nixos
<fresheyeball>
but usually there is a thing for it, that establishes those packages as a nix expression
<Fare>
do one — I'll adapt for the other ones.
<fresheyeball>
for example I use node2nix to manage npm packages inside of nix
schmittlauch[m] has joined #nixos
growpotk- has quit [Ping timeout: 246 seconds]
<fresheyeball>
it outputs a nix expression, with hashes that makes your npm packages pinned in nix, and reproducible and deterministic
Cale has joined #nixos
<fresheyeball>
then you NEVER run the npm command ever after
<fresheyeball>
opening the nix-shell for that sub project will symlink the npm packages from teh nix store
<fresheyeball>
and make them availabe to that shell
<fresheyeball>
the same expression can be used to build
<fresheyeball>
a nix-build does not have internet access, so you can't run the npm command inside your nix-build anyway
<fresheyeball>
this way you are guarenteed by nix that the packages you use for development in the nix-shell
<fresheyeball>
are the same being used in nix-build
<fresheyeball>
each "platform/language" in the nixpkgs repo has it's own tools for using that language with nix
<fresheyeball>
node has npm2nix and node2nix
knupfer has quit [Ping timeout: 264 seconds]
<fresheyeball>
there is stack2nix cabal2nix for haskell
<fresheyeball>
I don't know how the other ecosystems you need work, because I don't happen to use Ocaml or Lisp right now
<fresheyeball>
inside the nix-shell for your sub project, you can have incremental builds using standard tools
<fresheyeball>
I personally like to supply ALL my tools via the nix shell
<clever>
there is also yarn2nix for nodejs
<fresheyeball>
so that my version of cabal/stack is locked to the nix expression
<fresheyeball>
same for node, or whatever other tools you use in your project
<fresheyeball>
that way the nix-shell becomes a complete self contained dev environment where everything is pinned and deterministic, as well as identical to your build
<clever>
and baring weird stuff in your ~/.ghc/, it entirely gets rid of the "but it works for me" problem
jgeerds has quit [Ping timeout: 245 seconds]
chloekek has quit [Ping timeout: 246 seconds]
<clever>
somebody in here yesterday had something in ~/.ghc that was causing every single package from ghcWithPackages to be hidden
<fresheyeball>
Fare: you will also want to checkout gitignore source
<fresheyeball>
nix determines what to build based on if the source changed
<fresheyeball>
so you can get faster builds from the cache, if you change something like your README.md
trubydoor[m] has joined #nixos
mexisme has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
ixxie has quit [Ping timeout: 264 seconds]
<Fare>
fresheyeball, yeah, processing gitignore would be nice-- using src = /some/absolute/path is both not portable to another machine and includes stuff I'd rather ignore.
reallymemorable has quit [Quit: reallymemorable]
<Fare>
fresheyeball, can users still build with npm outside nix with node2nix ?
<fresheyeball>
Fare: yes
<fresheyeball>
but you should tell them to use the nix-shell
funrfy has joined #nixos
<Fare>
I admit I'm slightly worried about Windows users.
<clever>
a: windows users will need to continue to build things the old non-nix way
<mdash>
Fare: filterSource is an important function for those cases
<fresheyeball>
Fare: I run Nixos on my bare metal for dev
<fresheyeball>
but I know people who run nixos in a vm ontop of macos or windows
<fresheyeball>
and just ssh in, you get nice benefits over supporting a non-deterministic method in parallel
<fresheyeball>
mdash: regarding that actually
<fresheyeball>
is there a way to know what part of the source changed that is inspiring nix to rebuild?
<fresheyeball>
I have a build that is rebuilding unexpectedly, and I want to develop my debugging skill in preventing extranious rebuilds
glasserc has quit [Read error: Connection reset by peer]
<mdash>
fresheyeball: not directly, it just compares hashes
<mdash>
mrm, not sure if 'cd -' works like that non-interactively
<mdash>
heh mischan
selfsymmetric-mu has quit [Ping timeout: 245 seconds]
<evanjs>
worldofpeace: alright, just wasn't sure if it was appropriate to have the same dependencies listed in buildInputs _and_ in libPath (which is used in postFixup)
glasserc has joined #nixos
<symphorien>
fresheyeball: you can diff the two resulting store paths
<fresheyeball>
symphorien: what do you mean? I can see that the hash changed, but I don't know why!
<symphorien>
When nix uses a local direcotory it copies it in the store
<symphorien>
So you have the old directory and the new one somewhere in your store
<symphorien>
And you can diff them (with meld for example)
<fresheyeball>
symphorien: really?
<fresheyeball>
it caches the source directory? not just the output?
fendor has quit [Read error: Connection reset by peer]