Neo-- has quit [(Remote host closed the connection)]
<Infinisil>
Search for "PLUGGING" to find the points where i plugged it out/in
ibrahims has joined #nixos
Infinisil has quit [(Quit: leaving)]
<spinus>
looks like firmware is removed and not added again
ng0 has quit [(Quit: Alexa, when is the end of world?)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 opened pull request #27415: cc-wrapper: Always export environment variables for binutils (staging...cc-wrapper-prefix-binutils) https://git.io/vQdN7
NixOS_GitHub has left #nixos []
eacameron has joined #nixos
Infinisil has joined #nixos
celph has joined #nixos
<Infinisil>
No idea
<celph>
Git's triggering askpass whenever I try to push. I have key auth set up and I'd rather it just use SSH in the default way. I never explicitly asked for askPassword and I'm not sure how to tell Nix I'd rather not have it thanks.
<celph>
The options page isn't terribly informative on the subject
eacameron has quit [(Ping timeout: 260 seconds)]
<spinus>
Infinisil: looks like the firmware in unloaded
<spinus>
most stuff in unloaded and than loaded again, like pciehp
<spinus>
but fw-ohci is only unloaded (maybe it's loaded in the initrd and then unloaded and never loaded again?)
Infinisil has quit [(Ping timeout: 240 seconds)]
Infinisil has joined #nixos
<Infinisil>
No idea
<Infinisil>
Another display works no problem
<ottidmes>
Infinisil: I have no clue what I am talking about, but I saw a few iommu related errors and I read that it has to do with memory management for your hardware, could it maybe indirectly have to do with that? Or updating your BIOS?
<Infinisil>
ottidmes: Maybe, how would one update the BIOS? And I believe I'm using UEFI
pie_ has joined #nixos
<Infinisil>
I can dream of eventually being able to debug such problems properly, my internal linux knowledge is pretty minimal and therefore insufficient for this currently
<ottidmes>
celph: programs.ssh.askPassword = ""; will let git prompt you the password via the CLI, or do you mean never use password authentication with SSH, but always key based? Is that not a matter of setting up your SSH config correctly?
<celph>
ottimes: I use passwords for some hosts, but most of my regular ones I have keys for. I'm pretty sure it's configured correctly in that I've been able to remote into my server at work without issue
<Infinisil>
I do have updateMicrocode enabled in the configuration
<ottidmes>
Infinisil: The same for me, I just tend to scan the logs and hope that googling for scary messages finally yields some measure of success. Unfortunately the logs tend to have many scary bits that upon further investigation are nothing more than harmless warnings
<Infinisil>
ottidmes: Yeah, generally the logs are pretty calm, but on startup and device connections they turn wild
<celph>
ottidmes: I'm not sure why git's calling the GUI interface by default, I'll disable it and see if that fixes thigns
<Infinisil>
And btw the display works just fine all the time when I'm using macOS
<celph>
Infinisil: manual BIOS updates are usually less scary than they look but... I would rather not do them most of the time lol
<Infinisil>
Well, when I finally have nixos installed on my desktop pc (using a laptop now), I won't have to deal with the thunderbolt display anyways, as I could just use laptop for mobile, desktop with big screen at home
<Infinisil>
Right now it's laptop for development and desktop for gaming
<ottidmes>
Infinisil: That means it is Linux related and not BIOS/UEFI related, I guess. Did you already try to list the modules when it works (as soon as possible, maybe through some systemd service) and when it fails
<celph>
Weird, I rebuilt with programs.ssh.askPassword = ""; set but it's still opening ask password
<Infinisil>
ottidmes: lsmod outputs pretty much the same (bare some different numbers)
newhoggy has quit [(Remote host closed the connection)]
<Infinisil>
What I should really do is find out where xrandr (and therefore X11) gets its display information from and work from there
<ottidmes>
Infinisil: It started out for me like that, with my laptop and desktop, but now they are both so old already that dev work also shifted to the desktop and the laptop is just for writing (not heavy de) and remote access to my desktop
hellrazo1 has quit [(Read error: Connection reset by peer)]
<ottidmes>
Infinisil: Did you try changing your graphics driver?
<Infinisil>
ottidmes: Nope, how could this work?
hellrazo1 has joined #nixos
<celph>
Ah, it didn't change the environment variables within my shell session, that's why
<Infinisil>
ottidmes: My laptop is a MacBook Air from 2012, it's doing decent, but the battery life is horrible
<Infinisil>
Especially when a certain program *cough* firefox *cough* uses kinda a lot of power..
<Infinisil>
Damn you js
<Infinisil>
Seriously, why did a language written in 10 days take off to become the most used one..
<ottidmes>
Infinisil: My laptop ASUS from 2009 has a great battery life, just not enough CPU/MEM for my tastes, and I always use Firefox
ibrahims has quit [(Ping timeout: 268 seconds)]
<ottidmes>
Haha, that is insane indeed, something about critical mass
<Infinisil>
Imagine the world got somehow cought in a functional stream, where everyone was encouraged to learn Haskell and every imperative programmer would be laughed upon
eacameron has joined #nixos
<ottidmes>
I doubt it would be that much better, even do I prefer Haskell over most imperative programming languages, but you end up trading what set of problems for another set
<Infinisil>
ottidmes: What set of problems are you referring to?
schoppenhauer has quit [(Ping timeout: 260 seconds)]
<ottidmes>
Infinisil: Would this not let you use the fallback driver: services.xserver.videoDrivers = [ "vesa" ]
celph has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<Infinisil>
ottidmes: I should probably check if this driver is supported by my card first right?
<ottidmes>
Infinisil: "The driver auto-detects the presence of VESA-compatible hardware."
<ottidmes>
Infinisil: So I doubt it will fail to hard if it is not supported
<Infinisil>
Ah, okay I'll just try it
<Infinisil>
And I could just rollback anyways
<Infinisil>
restarting
thc202 has quit [(Ping timeout: 246 seconds)]
Infinisi1 has joined #nixos
tmaekawa has quit [(Read error: Connection reset by peer)]
<Infinisi1>
Nah, doesn't work
<Infinisi1>
x didn't start
Infinisil has quit [(Ping timeout: 248 seconds)]
tmaekawa has joined #nixos
justelex has joined #nixos
<Infinisi1>
oh well, I can't be bothered anymore, I'll look into it at some other time maybe
<ottidmes>
Infinisil: Regarding your question about what set of problems, in imperative programming you might have to deal with manual memory management, and thinking a lot more about how the machine will efficiently execute your instructions, where in e.g. Haskell you will run into problems with the performance of lazyiness and different kind of structuring problems
<Infinisi1>
ottidmes: Ah yes
radvendii has joined #nixos
<ottidmes>
Infinisi1: And efficient library level Haskell code is not what I call beatiful code
<Infinisi1>
Idris, which is really nice, is very similar to Haskell, but it's strict by default. It also has some unfinished linear type support, which could evolve into a rust-similar memory model. I feel like if more people would research stuff like this we could very well have an efficient purely functional language
justelex_ has quit [(Ping timeout: 260 seconds)]
slack1256 has joined #nixos
noobly has joined #nixos
<Infinisi1>
And such languages still allow for state via the state monad for example, it's just explicit all the time
<ottidmes>
Infinisi1: Yeah, I know about Idris, and about Coq and Agda as well. Idris definitely has some nice properties, but Haskell seems to also slowly progress towards more of a dependently typed language
<radvendii>
Infinisi1: are you saying the strict-ness is what makes it efficient?
<Infinisi1>
I eventually would like to contribute to idris, it's such a mind-blowing language, but sadly there's only one person working on it
<Infinisi1>
radvendii: Strictness can make it more efficient sometimes I think, but mainly it's more predictable
<radvendii>
Infinisi1: why does that make it more predictable? (not challenging, just trying to learn)
takle has quit [(Remote host closed the connection)]
<slack1256>
you can know how much memory the program has as you know how much memory each field of each data type occupies and you have the data at hand
takle has joined #nixos
<noobly>
Does anyone here use a any cryptocurrency wallets on NixOs other than electrum?
<slack1256>
on lazy evaluation, you have a closure that when evaluated correspond to the value. Meanwhile the closure can be bigger or smaller than the value itself and even when evaluation such closure, it could be constructed on such way it will need a lot of stack space for evaluations
<Infinisi1>
radvendii: I believe Haskell has often that problem: because of lazyness, you can never be sure when something actually gets executed, because it may only be needed much later, really hard to predict in a bigger program. And it may even be the case that a big calculation gets done twice, because it just calculates it lazily, without allocating memory for it
<slack1256>
which is emphatically not nice :-)
<Infinisi1>
slack1256: Good point
<radvendii>
Infinisi1: mm interesting
<radvendii>
but laziness is so much fun... </3
<slack1256>
although I consider myself on the camp that if the language is pure, then I would like it to be lazy :-)
<Infinisi1>
slack1256: Why so?
<radvendii>
same.
<radvendii>
i mean it allows a strictly greater subset of programs to evaluate
takle has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Quit: Leaving...)]
<slack1256>
I already know how to program to avoid most space leaks. Laziness' performance depends on both the producer and the consumer. I already know how to make good consumers
<Infinisi1>
radvendii: You mean infinite lists for example?
<slack1256>
And second I like that functions read like a bunch of definitions. That doesn't happen on a strict setting
<Infinisi1>
Well idris is just strict *by default*, there is a way to make things lazy explicitly
<radvendii>
Infinisi1: well for example `const 3 <infinite list>` won't evaluate in a strict language
lambdamu has joined #nixos
<radvendii>
there always is, but that means wrapping and unwrapping things, no?
<Infinisi1>
radvendii: In an *exclusively* strict language*
<Infinisi1>
radvendii: Idris handles it implicitly
<radvendii>
wait what? so it will figure out what won't ever get evaluated and not evaluate it?
<Infinisi1>
radvendii: Hold on, brb
Infinisi1 has quit [(Quit: leaving)]
Infinisil has joined #nixos
<Infinisil>
Alright I'll show you how it works in idris
<slack1256>
also, I don't like how laziness example always work with list. The real beauty is the composition of funcion operator
sigmundv_ has quit [(Ping timeout: 260 seconds)]
celph has joined #nixos
<Infinisil>
slack1256: Example?
<slack1256>
On a composition of functions which work on a recursive data type: `f . g . h` I like to think of f as the consumer and `g . h` as the producers.
<celph>
There, everything fixed for git
<ottidmes>
celph: What did you end up doing?
<slack1256>
I want f to dictate how to evaluate. For that it relies on bangpatterns and seq. Even better You can enforce it a data type level with bang pattern.
<celph>
The disabling worked, but only after rebooting. I also had kind of a minor stupid problem where my repo was configured for HTTPS and not SSH
<celph>
Now Everything is fine
<slack1256>
but for it really be efficient I need somthing of the consumers. Namely that they don't evaluate on their own, only define what should be in each field of the data. And second that their recursion proceed from the outer to the inner (also called guarder recursion)
<Infinisil>
radvendii: Ping, see the link i posted above
<slack1256>
which is why list example follow. They are an example of such pattern
<Infinisil>
slack1256: I see, yeah lazyness can be really useful
<ottidmes>
celph: It is often enough to just open a new terminal to reflect the enviroment changes, but good to hear it worked out!
<slack1256>
so if you really want modularity *at all cost* then laziness is a better default. As an incarnation of modularity is composition of functions
<slack1256>
just, define good consumers :-)
<slack1256>
which is what most serious haskell projects do in practice. See xmonad o irc-core. The most external data types are full of bang pattern in certain fields
<Infinisil>
slack1256: Some time ago I too was convinced that laziness is the one true way, but since reading a lot about Idris I changed my mind :P
<radvendii>
Infinisil: oh I see. so you just mark it at the type level in the function definition
<celph>
ottidmes: it didn't work that time, which is odd, but whatever, a reboot isn't a big deal
<ottidmes>
Semantics wise lazy or eager evaluation should not differ in Idris, since it is total, but I agree I do prefer lazy
<Infinisil>
radvendii: Yup, and since types are mandatore (at the top level), you always know if it's lazy or not
<slack1256>
Infinisil: heretic!
<slack1256>
:-)
<radvendii>
ottidmes: idris isn't lazy
<radvendii>
ottidmes: sorry strict*
<radvendii>
ahhh
<radvendii>
i mean total
<radvendii>
jeez
<radvendii>
my brain right now
<ottidmes>
radvendii: from there website "Idris is a total language"
<radvendii>
certain parts are total, but not general idris programs, right?
<radvendii>
oh really?
<ottidmes>
* their
<celph>
ottidmes: I thought briefly "well will this work if I log back out and back in again?" but didn't bother with it. It was easy enough to just reboot.
<radvendii>
huh. i stand corrected
<slack1256>
on a total language, recursion by definitions isn't unbounded. You work on recursive term with strictly reduced subparts
<Infinisil>
Idris programs don't *have* to be total
<slack1256>
but for it really to forgoe laziness, you need a concept of codata
<slack1256>
does idris have such concept?
<Infinisil>
But there is a totality checker, which can be used to error when functions aren't (for the whole file e.g.). Most of the standard library is total
<radvendii>
slack1256: I think it has codata
<slack1256>
laziness also murkies the difference between data and codata lol
<celph>
So that in short is why it wasn't using my SSH settings. Also, this probably has a quick solution, but when I start xrandr and I have a second monitor plugged into my laptop, it defaults to weird monitor settings. That might not be xrandr's problem and it's definitely just a config issue I've never bothered figuring out how to fix. When I toggle it off and on again it's fine but just that bit at startup
<ottidmes>
celph: I generally also do not bother, reboots are fast enough and then you can at least rule out it is not some stateful thing that is the cause of it still breaking
hyphon81 has joined #nixos
<slack1256>
Nice
<Infinisil>
Man I need to pick up my Idris book again where I left off, it's such a beautiful language and I'd love to master it
mog has quit [(Quit: im quiting!)]
<celph>
My configuration.nix has xrandrHeads = [ "HDMI1" "VGA2" "LVDS1" ], am I perhaps Doing It Wrong?
s33se_ has joined #nixos
<celph>
do I need to hyphenate them even though they aren't hyphenated in xrandr output?
<ottidmes>
A dependently typed language almost always is total, otherwise you get things like infinite loops in your type definitions
<Infinisil>
ottidmes: Ah yes, type-level stuff always has to be total
<slack1256>
which would mean that as a logic, is inconsistent
<ottidmes>
Exactly
<ottidmes>
What was it, types are theorems and functions are proofs?
Supersonic112 has quit [(Disconnected by services)]
<slack1256>
well, terms of data types are proof
Supersonic112_ has joined #nixos
<slack1256>
functions of how you generated them are also proof then
<celph>
I'll check if hyphenation works and then get back
Supersonic112_ is now known as Supersonic112
newhoggy has quit [(Remote host closed the connection)]
s33se has quit [(Ping timeout: 248 seconds)]
<slack1256>
for a proof to be valid, you need to reach the term in a finite amount of step
<slack1256>
clearly, infinitely recursive functions never reach a term, thus are invalid proof
mbrgm has quit [(Ping timeout: 260 seconds)]
<ottidmes>
slack1256: But unfortunately those finite steps can be so many, that it is practically infinite amount of steps
<slack1256>
we got certain pattern that we know are valid
<slack1256>
for example, on inductively defined data types (as the natural numbers)
radvendii has left #nixos ["WeeChat 1.9"]
<Infinisil>
ottidmes: The naturals can have pretty bad runtime when it's not optimized.. Try evaluating something like 7! using Nats in the interactive terminal
<slack1256>
if your function work in each recursive case in a subpart of you original data, that function is guarranted to terminate
mbrgm has joined #nixos
<slack1256>
terminate here is in a ideal computer
celph_ has joined #nixos
<slack1256>
not the shitty netbook from 2011 I am writing this on :-)
<celph_>
Update: hyphenation worked, but it only kind of worked
celph has quit [(Ping timeout: 260 seconds)]
<ottidmes>
Infinisil: I think I proofed that one, or something similar, as well a few years back, but then in Coq. Can't say I am fan
<celph_>
Or maybe it didn't work at all, I'm really not sure
magnetophon1 has joined #nixos
<celph_>
My output names are LVDS1, HDMI1, VGA1 (although sometimes it's assigned VGA2 because of reasons)
<celph_>
(Reasons I've never quite understood)
<ottidmes>
celph_: Is that setting necessary, I just run a script at boot that configures my multi monitor setup
<celph_>
ottidmes: I mean I easily could just run a script at boot, but I figured if I can do it in one line in configuration.nix then why bother
<Infinisil>
interactively evaluating 7! in Idris takes 5 seconds on my machine heh
<ottidmes>
celph_: I just let ARandR generate the script for me, so it is just as easy and I get consistent behavior that way, which you do not seem to get, so maybe worth a try?
<Infinisil>
Alright I'll head off, see ya
<ottidmes>
Infinisil: That was something I kept running into with Agda, when your proofs got complicated enough, it became way too slow
<ottidmes>
Goodbye, I am off too
<Infinisil>
ottidmes: With incremental compilation this wouldn't be a problem
<Infinisil>
wouldn't be as much of a problem*
<ottidmes>
Infinisil: But that is always the problem
<ottidmes>
Infinisil: In theory the compiler should be able to optimize such and such
<Infinisil>
ottidmes: Heh yes, the never-reachable optimal compiler
<ottidmes>
Infinisil: But it in practice it never seems to achieve that point
<Infinisil>
Does GHC have incremental compilation?
newhoggy has joined #nixos
<slack1256>
yes
<Infinisil>
Well that's nice to hear. I bet with enough effort and dedication Idris (and others) could get there too
<Infinisil>
GHC has received a lot of love
<ottidmes>
GHC also does a lot of aggresive inlining
<slack1256>
ghc got it easier. I got a clear separation of compile time (typechecking) and runtime (terms)
<celph_>
ottidmes: thanks for the suggestion, a question before you leave: anything special I have to call the script on boot?
<celph_>
wait, that's a stupid question
<slack1256>
idris got a more intermixed story by dependent typing, where types can depend on computation of terms and stuff
<ottidmes>
celph_: I just run it as part of my window manager setup script, as the very first thing
<slack1256>
thus the separation requeries a lot of analysis
<Infinisil>
slack1256: I don't have a deep idris knowledge, but this might not even be that big of a problem
<slack1256>
actually it is
<Infinisil>
oh
<slack1256>
that is the whole deal why depedent typing is so hard to catch mainstream
<slack1256>
you have to make the theorems go aways on the compiled result
<celph_>
ottidmes: I'm looking for the variable for that
<slack1256>
which is highly not trivial
<Infinisil>
slack1256: I see what you mean
<slack1256>
because the theorem themselv depend on the runtime
newhoggy has quit [(Ping timeout: 248 seconds)]
<slack1256>
what they are making progress on such front is really really good work
<celph_>
xserver.displayManager.sddm.setupScript?
<ottidmes>
celph_: That will depend on you setup, in my case using bspwm I have it run in my bsprc setup file
<Infinisil>
slack1256: Damn, this motivates me to learn for my formal methods exam which I'm gonna have in a couple weeks
<celph_>
ottidmes: I'm just using i3
<Infinisil>
Alright I'm actually off now, see ya
<ottidmes>
With i3 I used a session script, let me see
<slack1256>
me too, but thinking I would be tested make me lose motivation :-)
<celph_>
It looks like the sddm options are wm-agnostic
<celph_>
and dm-agnostic
<celph_>
So I will use them
<ottidmes>
celph_: If it works, by all means use that
<Infinisil>
slack1256: I'm so bad at preparing for tests, I haven't learned anything
<celph_>
ottidmes: definitely worth a try
Infinisil has quit [(Quit: I could talk for another hour if I won't go now)]
<slack1256>
I just can't muster the motivation to study /for a test/. I procastinate like crazy
newhoggy has joined #nixos
<slack1256>
I am currently doing it
<ottidmes>
celph_: In case it does not work, I used this: services.xserver.desktopManager.session = singleton { name = "custom"; start = "/path/to/your/arandr.sh"; } and services.xserver.desktopManager.default = "custom"
pie_ has quit [(Read error: Connection reset by peer)]
pie__ has quit [(Read error: Connection reset by peer)]
<ottidmes>
slack1256: I am sorry I have fed you more stuff to procrastinate with, that site got some great articles, I have to go now as well, goodbye!
<slack1256>
bye
<slack1256>
heh, so you see some irony. Cool
ottidmes has left #nixos ["WeeChat 1.7.1"]
<celph_>
ottidmes thanks, I'll give that a try if this doesn't work out
newhoggy has joined #nixos
noobly has quit [(Ping timeout: 260 seconds)]
magnetophon1 has quit [(Ping timeout: 248 seconds)]
<celph_>
I'm thinking this means I need to escape a $
newhoggy has joined #nixos
<slack1256>
missing ; ?
<celph_>
Is $$ the correct escape?
<celph_>
slack1256 nope
<celph_>
I'm embedding a script which uses a shell variable, I think it conflicts with nix' variable scheme
<celph_>
Recompiling with $$ seems to have worked but I'm not sure if it worked-worked or worked-but-did-the-wrong-thing worked
<celph_>
*rebuilding
newhoggy has quit [(Ping timeout: 240 seconds)]
<celph_>
welp, nothing to be done but try really. the manual doesn't specify escape characters so I might have just fucked things up but that's what nix is for
pie_ has quit [(Ping timeout: 240 seconds)]
justan0theruser has joined #nixos
cherrybl0ss0m_ has joined #nixos
celph_ has quit [(Ping timeout: 260 seconds)]
celph has joined #nixos
<celph>
It seems to have done nothing
justanotheruser has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
celph has quit [(Ping timeout: 260 seconds)]
celph has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
justan0theruser has quit [(Quit: WeeChat 1.7.1)]
justanotheruser has joined #nixos
<cherrybl0ss0m_>
I have a path in my Nix store that has an mtime of 1970-01-03 10:40:53.226881561 +0000. I am using a build node from a recent commit on master (within the last 7 days). When I try to build AMIs (on an EC2 instance) I get this error caused by this non-epoch mtime: `error: changing modification time of ‘/nix/store/00fhhmww8gmf6cg55b9f37fzp67wjm4l-diffutils-3.6’: Operation not permitted` Anyone seen this?
watersoul has joined #nixos
newhoggy has joined #nixos
boomshroom has joined #nixos
<boomshroom>
Hello. Been a while since logging in.
<boomshroom>
Is it possible to override fields of a derivation (namely src and version) without copying the whole derivation?
newhoggy has quit [(Ping timeout: 240 seconds)]
<boomshroom>
This is mostly to try a snapshot version of one package and to provide pending patches to another.
<boomshroom>
I do a naive // and the repl gives me the initial derivation. I send the merged set to a new stdenv.mkDerivation and a get a strange error.
<boomshroom>
`cannot coerce a set to a string, at (string):1:244`
celph has quit [(Ping timeout: 260 seconds)]
<cherrybl0ss0m_>
boomshroom: this might be the old way but you can do something like this `(terraform.overrideAttrs (oldAttrs: oldAttrs // { doCheck = false; }))`. I have that particular thing as an entry in the `buildIntputs` in a project shell.nix somewhere.
<cherrybl0ss0m_>
instead of doCheck you can set src or whatever you want.
newhoggy has joined #nixos
thoughtpolice has quit [(Ping timeout: 246 seconds)]
philips has quit [(Ping timeout: 255 seconds)]
yurrriq has quit [(Ping timeout: 255 seconds)]
mpickering has quit [(Ping timeout: 255 seconds)]
eddyb has quit [(Ping timeout: 258 seconds)]
gridaphobe has quit [(Ping timeout: 246 seconds)]
luto has quit [(Remote host closed the connection)]
dmj` has quit [(Remote host closed the connection)]
nand0p has quit [(Remote host closed the connection)]
peel has joined #nixos
<boomshroom>
cherrybl0ss0m_: I tried that along with overide and overrideDerivation and I get a new error:
<boomshroom>
error: attempt to call something which is not a function but a set, at /nix/store/5jc75g6f5zw7b1x9cd83ng57mh428hcs-nixos-17.09pre110213.01c3847b9c/nixos/pkgs/stdenv/generic/default.nix:348:52
newhoggy has quit [(Ping timeout: 268 seconds)]
peel has quit [(Client Quit)]
<boomshroom>
nvm: I didn't realise you were passing a lambda
celph has joined #nixos
<celph>
I'm getting error: undefined variable ‘singleton’ at /etc/nixos/configuration.nix:112:19
<boomshroom>
now that I have an expression that gives what I want, now to start a shell with it in the path. (I like doing things in nix-shell rather than simply installing them for some reason.)
newhoggy has joined #nixos
<celph>
I am very confused
<boomshroom>
celph: I'm guessing you're trying to access something called "singleton" on line 112 of your configuration.nix?
<celph>
Honestly I'm not even sure what singleton is, I'm just using the line ottidmes referenced to source a script
<celph>
boomshroom: 'In case it does not work, I used this: services.xserver.desktopManager.session = singleton { name = "custom"; start = "/path/to/your/arandr.sh"; } and services.xserver.desktopManager.default = "custom"'
<celph>
I can't find reference to singletons in the nixos manual or the nix manual
<boomshroom>
celph: Thank you. I was just about to ask you to post your configuration.nix on pastebin, but that should suffice. Would you mind sharing where you found that snipit?
<celph>
This chat, 10pm
newhoggy has quit [(Ping timeout: 258 seconds)]
<celph>
10:11 to be precise
<boomshroom>
Timezone? I can't check what was posted 1 hour from now.
<boomshroom>
celph: OK. It from what I can tell, you're trying to configure your monitor layout for i3. Lucky you, I also use i3 with multi monitors and my solution was exactly the same as with Arch Linux, which was to handle it in ~/.config/i3/config.
newhoggy has joined #nixos
slack1256 has quit [(Remote host closed the connection)]
<boomshroom>
The top of my i3 config has `exec xrandr ...` with most of the line being copied from "/path/to/your/arandr.sh."
<celph>
boomshroom: That's shockingly convenient. Although given it's i3 I'm not surprised.
<boomshroom>
The only thing I can think of that would make it more convenient is if I could find my arandr scripts. in configuration.nix, it's just setup to use i3 with "services.xserver.windowManager.i3.enable = true;"
newhoggy has quit [(Ping timeout: 260 seconds)]
<celph>
boomshroom: that's definitely possible with custom variables
<celph>
boomshroom: set a variable equal to your script and tell it to plop the script in your favourite directory
orivej has joined #nixos
<celph>
boomshroom: https://nixos.org/nix/manual/ I'm eventually hoping to make a complete system config and version it with git
<boomshroom>
celph: I meant that I should still have the scripts arandr generated for my Arch installation, but now I can't seem to find them.
<celph>
Oh lol sorry
<celph>
boomshroom: you wouldn't happen to know off the top of your head how to get LXDM to do it from boot too, would you?
<celph>
*startup
<boomshroom>
Ah ha! I found you .screenlayout! Sure enough the script is identical to the exec line in i3's config.
newhoggy has joined #nixos
vaibhavsagar has quit [(Remote host closed the connection)]
<boomshroom>
celph: From what I can tell looking at the options list, lxde and lxdm are not officially supported. I personally use sddm. It might take a little effort to use lxdm.
vaibhavsagar has joined #nixos
<celph>
Sorry it might be sddm, I'm not sure
<celph>
Just the default one, I haven't touched my dm settings
<boomshroom>
Closer look finds lxqt is available.
newhoggy has quit [(Ping timeout: 260 seconds)]
<celph>
boomshroom: it's SLiM
<boomshroom>
In that case it should be as simple as `services.xserver.displayManager.slim.enable = true; services.xserver.windowManager.i3.enable = true;` and then reboot. Assuming nixpkgs isn't doing something stupid, then SLiM should be able to find i3 just fine and your i3/config should be able to set the xrandr settings.
<celph>
boomshroom: yeah I'm just looking for pure aesthetics now, slim looks weird on multiple monitors with default settings lol
newhoggy has joined #nixos
<celph>
anyway gonna give it a try, might also switch to sddm so I can set a script in configuration.nix
<boomshroom>
celph: I'm not familiar with SLiM, so the rest is probably just checking SLiM's configuration options and adding desired ones to `services.xserver.displayManager.slim.extraConfig`.
<Ralith>
it is really distressing that nixos is *still* using an abandoned display manager
<celph>
boomshroom: thanks
<Ralith>
for the default
<celph>
boomshroom: I don't feel to strongly about my DM so I may also just switch to something less barebones
<boomshroom>
SDDM just displays the same screen on both displays. (not from the same buffer as each screen has its own password prompt). It is a little akward that SDDM thinks my screens are swapped.
<boomshroom>
Something about how primary monitors are assumed to be to the left, while mine is to the right.
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
<boomshroom>
Hmm... I got a overriden version of vivaldi, but when I try to use it, it reports its version as 1.10 instead of 1.11.
newhoggy has quit [(Ping timeout: 240 seconds)]
Sonarpulse has quit [(Ping timeout: 240 seconds)]
vaibhavsagar has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<boomshroom>
Well, goodnight!
boomshroom has left #nixos []
newhoggy has quit [(Ping timeout: 268 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
vaibhavsagar has joined #nixos
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
ebzzry has quit [(Ping timeout: 260 seconds)]
ebzzry has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
joehh has joined #nixos
<joehh>
hi, is there a standard way to generate a nixos vm image containing custom packages?
Ivanych has joined #nixos
<joehh>
ideally an ova file
newhoggy has joined #nixos
leat has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
celph has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
orivej has quit [(Remote host closed the connection)]
orivej has joined #nixos
vincent_vdk has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
vincent_vdk has joined #nixos
peel has joined #nixos
newhoggy has joined #nixos
vincent_vdk has quit [(Client Quit)]
vincent_vdk has joined #nixos
Ivanych has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 240 seconds)]
wowi42 has quit [(Ping timeout: 240 seconds)]
peel has quit [(Ping timeout: 240 seconds)]
<srhb>
joehh: I think there used to be at least. Try grepping for ova in nixos
<srhb>
Is there a shortcut for duplicating a part of the arguments into buildInputs? Lots of duplication...
<NixOS_GitHub>
[nixpkgs] elitak opened pull request #27419: oraclejdk8: armv7l support (master...armjdk) https://git.io/vQFIW
NixOS_GitHub has left #nixos []
<srhb>
In which phase should python files have their patches shebanged, if they're going into $out/bin ?
<srhb>
... their shebanges patched
* srhb
gets coffee
newhoggy has quit [(Ping timeout: 246 seconds)]
<srhb>
I suppoes postBuild would make sense
iyzsong has quit [(Read error: Connection reset by peer)]
iyzsong has joined #nixos
vaibhavsagar_ has joined #nixos
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
<joehh>
looks like nixos/lib/make-disk-image.nix is the key bit
<srhb>
Mic92: ping. I see you've upversioned lttng-ust. I think the shebang in bin is wrong (/usr/bin/env python) - but I'm not sure how it should preferably be fixed. Should I just add a python propagatedBuildInput?
<srhb>
joehh: Yeah :)
vaibhavsagar_ has quit [(Ping timeout: 260 seconds)]
orivej has quit [(Ping timeout: 260 seconds)]
orivej has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
justbeingglad has joined #nixos
justbeingglad has left #nixos []
newhoggy has joined #nixos
zeus_ has joined #nixos
newhoggy has quit [(Ping timeout: 268 seconds)]
<srhb>
We need more euro early birds in this channel to support my nosey questions... :-)
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 268 seconds)]
kier has joined #nixos
newhoggy has joined #nixos
<joehh>
or ausie experts :)
<Mic92>
srhb: well it's 9:00 am
<Mic92>
not too early
<Mic92>
python must be in buildInputs. so that the fixupPhase, replace the shebang correctly
newhoggy has quit [(Ping timeout: 246 seconds)]
Wizek has joined #nixos
newhoggy has joined #nixos
Wizek_ has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vQFtq
<srhb>
Mic92: Ah, yes, that's what I ended up doing :)
<srhb>
I was just a little confused of whether it should be propagatedBuildInputs because a dependent package needs to find that package too, but I guess not since the path will be fixed to the store path.
<joepie91>
[21:07] <gchristensen> joepie91: it is super hard to do what you're suggesting
<joepie91>
which aspect specifically? I might be able to help
MinceR has quit [(Ping timeout: 260 seconds)]
thc202 has joined #nixos
MinceR has joined #nixos
joehh has joined #nixos
ibrahims has joined #nixos
newhoggy has joined #nixos
<LnL>
what was your idea again?
hyphon81 has quit [(Remote host closed the connection)]
<joepie91>
LnL: was that question directed at me? :P
<LnL>
yes :)
newhoggy has quit [(Ping timeout: 240 seconds)]
<Mic92>
srhb: shebangs are a runtime dependencies. For these case we do not propagtedBuildInputs
<joepie91>
LnL: depends on what they were referring to; my most notable two suggestions towards gchristensen were 1) put docs elsewhere, not in the code itself, and 2) start by writing a function reference and only then writing guide-style documentation on top of that, referencing the function reference items (as a response to him showing me a work in progress that I felt was confusing because it tried to be both at the same time)
<clever>
Mic92: i would generally prefer to run patchSheBangs over the dir the script is in, so it just works all the time
<Mic92>
clever: this is what I did in the end
simukis has joined #nixos
joepie91 is now known as not-a-titlebot
not-a-titlebot is now known as joepie91
newhoggy has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] domenkozar closed pull request #1467: use sha256 hashes in the examples (master...nomd5) https://git.io/vQdPF
NixOS_GitHub has left #nixos []
<clever>
LnL: ah, it was out of disk space yesterday
joepie91 is now known as joepie91__
<LnL>
that's probably it then
joepie91__ is now known as joepie91
<LnL>
I wonder why tho, both builds and the build logs are streamed streight to s3
<clever>
there still is the postgres db
<clever>
and the journal
<clever>
but those shouldnt be that big...
newhoggy has quit [(Ping timeout: 240 seconds)]
hotfuzz_ has joined #nixos
Supersonic112 has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vQFmM
<NixOS_GitHub>
nixpkgs/master e8a8f12 Christian Kögler: snapper: add nixos module...
mpcsh has quit [(Quit: THE NUMERICONS! THEY'RE ATTACKING!)]
zeus_ has quit [(Remote host closed the connection)]
mpcsh has joined #nixos
zeus_ has joined #nixos
zeus_ has quit [(Ping timeout: 246 seconds)]
ixxie has quit [(Quit: Lost terminal)]
ixxie has joined #nixos
Filystyn has joined #nixos
<gchristensen>
joepie91: I suppose it feels difficult to make a function reference without diving in to a narative about it
DutchWolfie has joined #nixos
<Filystyn>
so how do i reset root pass in nix
<Filystyn>
grub trick not working
ng0 has joined #nixos
pie_ has joined #nixos
jbgi has joined #nixos
<joepie91>
gchristensen: interesting, people usually have the opposite problem :)
<joepie91>
gchristensen: can you give an example of a concrete issue you ran into while trying to create a function reference? like, a particular point that made you go "hm, I'm not sure what to put here / how to approach this / etc."?
jbgi has quit [(Client Quit)]
jbgi has joined #nixos
predkambrij has joined #nixos
Ivanych has joined #nixos
<jbgi>
about channels for nix-env, it looks like nixos-unstable is more up-to-date than nixpkgs-unstable. I was expecting the opposite, as nixpkgs-unstable is a subset of nixos-unstable. Any idea why? alternative question: is there any point in using nixpkgs-unstable instead of nixos-unstable?
<gchristensen>
jbgi: sometimes one goes faster than the other
<gchristensen>
jbgi: it is because of different requirements. if you're using it as an OS, you want nixos-unstable. nixpks-unstable however won't update unless darwin packages build
<joepie91>
gchristensen: the function name list here is a bit messed up due to the CSS that npmjs.org uses, but this is an example of how I approach a function reference (without dedicated tooling): https://www.npmjs.com/package/bhttp#api
<joepie91>
so if you ignore the unreasonable amount of whitespace between the get/post/etc. function signatures, it will give a good idea
<joepie91>
:p
<gchristensen>
ah
ng0 has quit [(Quit: Alexa, when is the end of world?)]
<joepie91>
gchristensen: if you have a bit more freedom in styling (which docbook should give you), you could pretty much just follow a hierarchy along the lines of: category -> function -> arguments
<joepie91>
# buildEnv
<joepie91>
## pname
<joepie91>
description goes here
<joepie91>
etc.
<joepie91>
I'm not sure of the capabilities of docbook so you'll want to experiment a bit with eg. the function signature representation
<gchristensen>
well am I not doing that already?
<joepie91>
in JS it's convention to mark optional arguments with [] and to show `options` arguments as just `options` in the signature (not listing the options) and expanding upon them in the body, but this doesn't map very cleanly to Nix where there's only single arguments and attribute sets
<joepie91>
gchristensen: in part
<joepie91>
gchristensen: you don't have a clear header for bundlerEnv for example, and you have examples inline with the argument definitions
<gchristensen>
you asked for a specific example of where I'm having trouble. can I turn that around and ask for a specific example of something I've written, and how it might be better?
<ij>
I've only applications defined in my configuration.nix, but some package's've put a lot of files into /etc. Does nixos give you some packages by default?
<gchristensen>
oh that makes sense
<joepie91>
if I were to skim the current version of your documentation, I'd interpret that as "there's a thing Ruby and it has properties "pname", "name", etc."
<gchristensen>
oh I see
<joepie91>
not as "there's a language Ruby, and there's a function "bundlerEnv", and that function has arguments "pname", "name", etc."
<gchristensen>
yes
<joepie91>
basically, you'd want the function reference to be very structured, methodical, almost... clinically consistent?
<Filystyn>
so how do i reset root password
<gchristensen>
well that is a trivial fix
<gchristensen>
ij: what sort of files?
<gchristensen>
Filystyn: do you have access to sudo/
<joepie91>
gchristensen: it's a bit difficult for me to give concrete feedback on this particular bit of documentation, content-wise, because I am not familiar with Ruby and its ecosystem :P
<joepie91>
so I don't have a very solid frame of reference
<gchristensen>
joepie91: how does the header of of buildRubyGem look?
ixxie has quit [(Ping timeout: 240 seconds)]
<ij>
Scratch that, I'm not sure this is the best way to get at my problem.
<gchristensen>
ij: what is your problem?
<joepie91>
gchristensen: that's a fine header, but I'm not sure it should be represented as a numbered section as such, although that may be a bit of a docbook limitation -- it's difficult to distinguish it from non-reference documentation right now
takle has joined #nixos
<gchristensen>
yeah
<joepie91>
does docbook automatically number things?
<aristid>
gchristensen: nixos-unstable is ahead of nixos-unstable-small?
butchery has joined #nixos
<gchristensen>
it does, joepie91
<joepie91>
ah, that's a bit annoying
<joepie91>
gchristensen: very loosely related, I've been experimenting in the past with language-agnostic declaratively specified function references: https://gist.github.com/joepie91/fe7651c96d80eb9511e508123091060a -- it certainly requires more thought on how to represent things in Nix, but it's an example of providing structured documentation independently from the code files themselves (but with the ability to correlate it to the code)
<gchristensen>
huh
butchery has quit [(Client Quit)]
mkoenig has joined #nixos
<joepie91>
gchristensen: such an approach would allow both documentation generation as well as integration with eg. editor autocomplete, without cluttering up source files
butchery has joined #nixos
<joepie91>
(as well as automatic "is this new thing documented" analysis for eg. PRs)
<joepie91>
not-super-remarkable-but-nevertheless-nice bonus: it allows for working with the documentation without necessarily needing a full-fledged language parser
<joepie91>
it's standard YAML, and there's YAML parsers for pretty much everything
nslqqq has quit [(Ping timeout: 268 seconds)]
<ij>
In my docker build the locale isn't properly set(at all). Why did I get proper locale in nixos, even though I haven't defined anything related in configuration.nix?
nslqqq has joined #nixos
<gchristensen>
ij: how did you make the docker image?
<bennofs>
gchristensen: may even need to logout/login again?
<gchristensen>
yeah maybe
<bennofs>
I've always wondered why we do this impure thing for aspell but for nothing else
<gchristensen>
same!
<gchristensen>
ugly
<Eisfreak7>
New terminal doesn't work (I installed aspellDics.en but (neo)-vim still prompts me to install "en" in utf-8)
<Eisfreak7>
I'll try relogin
Filystyn has joined #nixos
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
<Filystyn>
srhb
<srhb>
Filystyn: Yes?
<Filystyn>
i did it i set the new root password and after logging in the password fails to authenticate
<srhb>
Filystyn: How did you do it exactly?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth closed pull request #27398: tinc: allow the daemon to write to files in /etc/tinc/${network}/hosts (master...tinc-fix) https://git.io/vQdsR
NixOS_GitHub has left #nixos []
<Filystyn>
mount sda1 go to /mnt nixos-install (set the root)
<Eisfreak7>
bennofs gchristensen: Still not working :( `ASPELL_CONF` is `~/.nis-profile/lib/aspell` which contains `en.multi`, but nvim doesn't find it
<srhb>
Filystyn: What do you mena "go to /mnt nixos-install"
<Filystyn>
well i just did cd /mnt
<srhb>
Filystyn: You need to use nixos-install --chroot
<Filystyn>
i checked i am on my system dir
<bennofs>
Eisfreak7: what does `aspell dicts` say
<srhb>
Filystyn: That will change your root to /mnt
<srhb>
Filystyn: Only then will passwd change the right file
<Filystyn>
ooh
<srhb>
Filystyn: Otherwise you're just changing the (in-memory) installer passwd
michaelpj has quit [(Read error: Connection reset by peer)]
<Filystyn>
but man somethign actualy did install
<Filystyn>
OK
<Filystyn>
brb
Filystyn has quit [(Client Quit)]
<srhb>
and.. nevermind
<Eisfreak7>
bennofs: That aspell isn't installed (is it needed?). After installing it litsts multiple dics, among others `en`. I'm unsing `:set spelllang=en` and `set spell`.
sary has quit [(Ping timeout: 240 seconds)]
sary has joined #nixos
<ij>
Judging by all-packages.nix, dockerTools are defined in pkgs/build-support/docker, right?
butchery has quit [(Quit: Lost terminal)]
<bennofs>
ij: yes
<bennofs>
Eisfreak7: looks like vim has builtin spellchecking that does not use any of the standard spellchecking backends
<Eisfreak7>
bennofs: So there's nothing I can do besides letting vim fetch the dicts itself?
pie_ has quit [(Ping timeout: 260 seconds)]
pie_ has joined #nixos
Filystyn has joined #nixos
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
<Filystyn>
srhb it worked thank you
<srhb>
Filystyn: Great :)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm pushed 1 new commit to master: https://git.io/vQFni
<NixOS_GitHub>
nixpkgs/master 96b4ed6 Eric Litak: factorio: 0.15.26 -> 0.15.30 (#27421)
NixOS_GitHub has left #nixos []
<Filystyn>
what wa sthe command to remove old nixos versions?
ixxie has quit [(Quit: Lost terminal)]
<Filystyn>
so no rollbacks for now
michaelpj has joined #nixos
mog has joined #nixos
<necronian>
How would I set my loginShell in ldap? I naively set it to /bin/bash, but I quickly realized when I tried to ssh in that it wasn't going to work.
<Filystyn>
k i think i gotit
<necronian>
I get User testuser not allowed because shell /bin/bash does not exist
<ij>
bennofs, Does the nix/nixkgs manual have anything about what modules are? (Like the i18n module.)
<bennofs>
Eisfreak7: stat("/home/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/wrappers/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.local/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/etc/per-user-pkgs/bennofs/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.nix-profile/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/var/nix/profiles/default/bin/curl", 0x7ffffed5b200) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/store/sklr65vffwynr08amw4xw2i09laylacy-gnome-breeze/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/opengl-driver/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<Eisfreak7>
bennofs: Thanks for the detailed writeup :D
<bennofs>
stat("/run/opengl-driver-32/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/etc/per-user-pkgs/bennofs/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.nix-profile/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/var/nix/profiles/default/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/current-system/sw/share/nvim/site/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/store/whyjkr43gm1k286jhsbhryi0vvd5p8v5-neovim-0.2.0/share/nvim/runtime/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/current-system/sw/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/var/nix/profiles/default/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.nix-profile/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/etc/per-user-pkgs/bennofs/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/opengl-driver-32/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/opengl-driver/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/store/sklr65vffwynr08amw4xw2i09laylacy-gnome-breeze/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
newhoggy has joined #nixos
<bennofs>
stat("/home/.local/share/nvim/site/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/run/current-system/sw/etc/xdg/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/nix/var/nix/profiles/default/etc/xdg/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.nix-profile/etc/xdg/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<srhb>
With a proper irc logging service, you could store entire filesystems like this!
<bennofs>
stat("/etc/per-user-pkgs/bennofs/etc/xdg/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/etc/xdg/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.config/nvim/plugged/rust.vim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.config/nvim/plugged/vim-coffee-script/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<bennofs>
stat("/home/.config/nvim/after/spell", 0x7ffffed58d40) = -1 ENOENT (No such file or directory)
<joepie91>
gchristensen: that's based on a sliiightly modified version of that YAML I pasted btw
<Filystyn>
do i need to enable tor channel on nix?
<Filystyn>
configuratioN8
<Filystyn>
to use tor browser
<Filystyn>
seems it can not send recv
<bennofs>
it solves the problem of units in systemd-user session not having access to the normal environment variables, but without polluting the whole systemd-user instance with environment variables (as systemctl --user import-environment would do)
newhoggy has quit [(Ping timeout: 260 seconds)]
<bennofs>
and it can be extended to support other envs in the same systemd instance (for example, if you want to launch graphical programs, you'd have a init-graphical target etc)
<srhb>
Woot, ceph luminous built! It only took me one and a half day. :-P
<srhb>
I need a build farm...
<necronian>
Okay now I want to use sssd+ldap for sudo, but I need to add sudoers: files sss to /etc/nsswitch.conf Is there a way for me to just append a string to the end of that file?
newhoggy has joined #nixos
lvernschrock has joined #nixos
<srhb>
necronian: Can you modify environment.etc."nsswitch.conf".text ?
ibrahims has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
<qknight>
if i have type = types.submodule to add default values to a `set` of values. how can i apply this to a `set` which i create by using `import`
<necronian>
srhb: Maybe, I have no idea how though, I tried environment.etc."nsswitch.conf".text = ${environment.etc."nsswitch.conf".text} + "\nsudoers: file sss";
<bennofs>
necronian: try just environment.etc."nsswitch.conf".text = "sudoers: file sss";
<necronian>
bennofs: Oh that worked, so append is the default?
<necronian>
I assumed I would only end up with a file with just that line in it
<bennofs>
necronian: it depends on the configuration option type. but seems it is for .text (.text has type: types.lines)
<srhb>
I think any kind of recursive attempt is going to go badly. modules don't have a good override story.
newhoggy has joined #nixos
ixxie has joined #nixos
newhoggy has quit [(Ping timeout: 268 seconds)]
Turion has joined #nixos
jophish has quit [(Ping timeout: 240 seconds)]
<Turion>
How does ssh agent work in nixos? By default, programs.ssh.startAgent is true, but still I have to enter my ssh password every time, what am I doing wrong?
<ixxie>
Say I want to install hydra in tandem with nixops and disnix, to setup a CI microservice arch.: can I hook it up with hydra.nixos.org in order to take advantage of its cache?
jophish has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] elitak opened pull request #27422: factorio: authenticatedFetch now handles 404 properly (master...factorio-version) https://git.io/vQFWL
NixOS_GitHub has left #nixos []
Neo-- has quit [(Remote host closed the connection)]
Neo-- has joined #nixos
<NickHu>
Is anyone else finding virtualbox not working in latest nixos-unstable?
lvernschrock has quit [(Ping timeout: 248 seconds)]
radvendii has joined #nixos
<radvendii>
is there a package for natpmp? I can't find one under the name 'natpmp' but maybe it's under a different name?
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
<LnL>
srhb: you can disable a module with unstable, but it's not an ideal solution
mudri has quit [(Ping timeout: 248 seconds)]
<srhb>
Is there a common solution to buildPythonPackagescripts failing when setup.py specifies a version constraint (like Cheetah>=2.0.1) despite the propagatedBuildInput cheetah being 2.4.4 ?
newhoggy has joined #nixos
<srhb>
I feel like it should just work.
<samae>
aszlig: Do you use the taskserver module with your own CA?
<srhb>
Could not find a version that satisfies the requirement Cheetah>=2.0.1 (from test==1) (from versions: )
<srhb>
Is it suspect that the versions list is empty?
ibrahims has quit [(Ping timeout: 248 seconds)]
cpennington has quit [(Remote host closed the connection)]
zeus_ has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
<srhb>
If I remove the version constraint entirely, it works. O_o
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Nadrieril opened pull request #27423: tinc service: BindToAddress and ListenAddress are different options, they should not be mistaken (master...fix-tinc-listenaddress) https://git.io/vQFlY
NixOS_GitHub has left #nixos []
<Filystyn>
do i need to enable tor in configuration.nix ?
covux has joined #nixos
<covux>
hi. iam trying to use vst-plugin Diva from u-he (via bitwig-studio) and it seems that some dependencies missing. it crashes when i open the register- or preset-dialog. output of ldd https://pastebin.com/ecmrQe9g how can i provide the missing libs?
zeus_ has quit [(Ping timeout: 246 seconds)]
newhoggy has joined #nixos
<ottidmes>
radvendii: There does not seem to be a package for it, only support for it in miniupnpd and filegive. Based on Arch Linux's package for it, it does seem simple enough to package: https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=natpmp
<radvendii>
ottidmes: cool, thanks. I'm not sure I need it, but maybe I'll package it anyways. seems useful to have around
newhoggy has quit [(Ping timeout: 248 seconds)]
ibrahims has joined #nixos
<Filystyn>
I won't drop with this tor problem it's definatly something I miss in configuration.nix
<covux>
ottidmes: thanks. i give it a try. wrapper script is not possible imho since its a plugin, which is loaded by another executable
newhoggy has joined #nixos
cpennington has joined #nixos
mkoenig has quit [(Remote host closed the connection)]
ixxie has quit [(Quit: Lost terminal)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg opened pull request #27424: stdenv: Don't put man pages into `doc` outputs (staging...no-man-in-doc) https://git.io/vQF43
NixOS_GitHub has left #nixos []
mkoenig has joined #nixos
newhoggy has quit [(Ping timeout: 268 seconds)]
cpennington has quit [(Remote host closed the connection)]
<joepie91>
gchristensen: I'll whip up an example for Nix-y stuff Some Time Soon
<joepie91>
lost my focus for now :p
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
<NickHu>
Does command-not-found work for anyone else?
<NickHu>
I recall seeing a github issue where they said it was broken but I can't find it anymore
<bennofs>
NickHu: what channel are you on
hyphon81 has joined #nixos
<NickHu>
nixos-unstable
<NickHu>
It's possible that I might just not have it installed - I could've sworn it worked by default back on 16.09 stable with no extra config though
Neo-- has quit [(Remote host closed the connection)]
bugworm has quit [(Quit: Konversation terminated!)]
<Filystyn>
welp definatly tor browser is broken on nix
bugworm has joined #nixos
bugworm has quit [(Client Quit)]
peel has joined #nixos
<Filystyn>
ok sorry for the fuss this is tor it self problem
peel has quit [(Client Quit)]
ixxie has joined #nixos
newhoggy has joined #nixos
Ivanych1 has quit [(Ping timeout: 268 seconds)]
bennofs has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 246 seconds)]
Eisfreak7 has quit [(Ping timeout: 268 seconds)]
Eisfreak7 has joined #nixos
<bennofs[m]>
I'd really like to release nix-index and add it to nixpkgs. But I don't feel comfortable until https://github.com/bennofs/nix-index/issues/11 is solved. Does someone have an idea how to solve / c workaround this?
newhoggy has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 3 new commits to staging: https://git.io/vQFRn
<NixOS_GitHub>
nixpkgs/staging 093cc00 John Ericson: cc-wrapper: Always export environment variables for binutils...
<NixOS_GitHub>
nixpkgs/staging 3739858 John Ericson: cc-wrapper: Use new bash uppercase subsitution syntax in setup hook...
<NixOS_GitHub>
nixpkgs/staging 2113631 John Ericson: Merge pull request #27415 from obsidiansystems/cc-wrapper-prefix-binutils...
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to staging: https://git.io/vQFR4
<NixOS_GitHub>
nixpkgs/staging ae26f29 Robin Gloster: systemd: 233 -> 234
NixOS_GitHub has left #nixos []
newhoggy has quit [(Ping timeout: 248 seconds)]
pie_ has joined #nixos
newhoggy has joined #nixos
ng0 has joined #nixos
ng0 has quit [(Client Quit)]
newhoggy has quit [(Ping timeout: 240 seconds)]
pie_ has quit [(Read error: Connection reset by peer)]
pie_ has joined #nixos
pie__ has joined #nixos
pie_ has quit [(Client Quit)]
[0__0] has quit [(Ping timeout: 268 seconds)]
mudri has joined #nixos
[0x4A6F] has quit [(Remote host closed the connection)]
magnetophon has joined #nixos
newhoggy has joined #nixos
ixxie has quit [(Quit: leaving)]
zeus_ has joined #nixos
ixxie has joined #nixos
ixxie has quit [(Client Quit)]
newhoggy has quit [(Ping timeout: 258 seconds)]
ixxie has joined #nixos
zeus_ has quit [(Ping timeout: 258 seconds)]
stanibanani has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
newhoggy has joined #nixos
Infinisil has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
magnetophon has quit [(Read error: Connection reset by peer)]
newhoggy has joined #nixos
Neo-- has quit [(Remote host closed the connection)]
jonte has quit [(Ping timeout: 240 seconds)]
Neo-- has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
<cherrybl0ss0m_>
I have a path in my Nix store that has an mtime of 1970-01-03 10:40:53.226881561 +0000. This is not the UNIX EPOCH. How do I most cleanly rectify my current /nix/store that contains files like this without hacking stuff up too much outside of the norms of the Nix store operationally?
<cherrybl0ss0m_>
I can do it through brute force with a command run as root, but perhaps there is a `nix-store` option that does this that I am missing in the docs?
<cherrybl0ss0m_>
(I was going to `unshare -m bash` and then in the root shell after remounting /nix/store as rw I would "do gross stuff to the mtime" manually, which I would rather not do.)
<clever>
cherrybl0ss0m_: i think its safe to just leave the mtime like that
<clever>
which file exactly has that time?
newhoggy has joined #nixos
<cherrybl0ss0m_>
clever: I get errors when building EC2 AMIs via the nixpkgs create-amis.sh because of the mtime discrepancy.
<cherrybl0ss0m_>
So this impedes my progress and violates the notion of deterministic builds, I would suspect?
<clever>
ah, what is the error?
<samae>
there's the `nix-store --verify --repair` but it may be quite an overkill
<cherrybl0ss0m_>
`error: changing modification time of ‘/nix/store/00fhhmww8gmf6cg55b9f37fzp67wjm4l-diffutils-3.6’: Operation not permitted`
<cherrybl0ss0m_>
samae: that did not change the mtime for me in my nix store, though I tried that.
<clever>
can you give more context for the error?
<samae>
ok
<clever>
it shouldnt be trying to change things i think
<aszlig>
samae: to be honest, i don't even use the taskserver module at all at the moment
<samae>
aszlig: ok, because I'm having a hard time tracing with I have this error
<samae>
tracing why, sorry
<cherrybl0ss0m_>
in a nixpkgs.git clone locally in this NixOS EC2 instance I cd into the nixpkgs clone and run: `./nixos/maintainers/scripts/ec2/create-amis.sh`
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] joachifm pushed 2 new commits to master: https://git.io/vQFzf
<NixOS_GitHub>
nixpkgs/master 9953a79 Eric Litak: factorio: authenticatedFetch now handles 404 properly
<NixOS_GitHub>
nixpkgs/master f752b15 Joachim F: Merge pull request #27422 from elitak/factorio-version...
NixOS_GitHub has left #nixos []
lvernschrock has joined #nixos
<clever>
cherrybl0ss0m_: can you gist more of the error?
<aszlig>
samae: which error exactly?
<samae>
“Error: CA setup not done or incomplete” complaining it cannot find ca.key
<clever>
so if root is a member, the system isnt going to survive
vaibhavsagar has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
<cherrybl0ss0m_>
as mentioned above that isn't the error, that's a warning and even when I fixed locally (I would need to build an AMI locally and create a new instance and it takes an our for me to upload right now, so I am not doing it) but on the EC2 instance I get the additional mtime error as you see above
<cherrybl0ss0m_>
that is the actual error in this case that is preventing the AMI from being built.
<clever>
the error is happening inside the part of nix that ensures everything has an mtime of 1
<cherrybl0ss0m_>
I have the same warning that still builds the AMI outside of EC2
<aszlig>
samae: you get that error at runtime, right?
<samae>
aszlig: yes
<clever>
cherrybl0ss0m_: so the problem is actualy when nix is trying to enforce what your expecting it to enforce
<aszlig>
samae: ah wait...
<clever>
cherrybl0ss0m_: what user owns /nix/store/ on that system?
<aszlig>
samae: found the bug, fixing...
<samae>
oh, nice! that was quick
<cherrybl0ss0m_>
clever: the mtime is being modified inside of the same fake chroot thing, that copumpkin added a couple of months ago for image building (to speed it up)
<clever>
cherrybl0ss0m_: ahh
<cherrybl0ss0m_>
that is where the problem arises
<aszlig>
samae: the fix however will take ab it, because i need to add a test for manual config
<cherrybl0ss0m_>
but I question why my files in the nix store are set to a different value to begin with
nh2 has quit [(Quit: Leaving.)]
<clever>
cherrybl0ss0m_: what does this command say: nix-store -q --roots /nix/store/00fhhmww8gmf6cg55b9f37fzp67wjm4l-diffutils-3.6
<samae>
ok, but nice anyway
<samae>
aszlig: anything I can change right now to make it start here? Or should I postpone this task
<cherrybl0ss0m_>
It feels like there is something wrong happening well before the chroot part of the image building process?
<clever>
cherrybl0ss0m_: yeah, thats a possibility
simukis has quit [(Remote host closed the connection)]
newhoggy has quit [(Ping timeout: 255 seconds)]
Itkovian has joined #nixos
[0__0] has joined #nixos
newhoggy has joined #nixos
<S0rin>
Hey I am wondering is it possible to work with multiple channels like stable / unstable ?
lambdael has quit [(Quit: WeeChat 1.7.1)]
<S0rin>
On my system I mostly can run on stable but there some package that are too old, or just not present
<ottidmes>
S0rin: You can, but you might end up with two versions of many packages, but e.g. a few packages with not that many dependencies themselves is perfectly viable
ryanartecona has joined #nixos
<ottidmes>
S0rin: I have that as well, mostly stable 17.03 and a few from unstable that are missing or to outdated
<S0rin>
As I started in nixos I switched to unstable then to master (with the -I ~/nixpkgs option) but master is a bit too much frequently broken to y need
<clever>
gchristensen: strange, no eval has passed in the last 21 hours
crst has joined #nixos
<gchristensen>
clever: out of space :((
<S0rin>
ottidmes: oh fine can you point me to some doc or explain a bit the how to ?
<clever>
gchristensen: ah, so it wasnt really fixed
crst is now known as Guest73314
<simpson>
S0rin: Folks who are developing on nixpkgs usually only use master for the specific things that they are developing, and otherwise stick to a channel for the rest of the system.
<gchristensen>
oh I might be wrong
newhoggy has quit [(Ping timeout: 260 seconds)]
<clever>
gchristensen: the postgres was dead, but that came up just recently
<S0rin>
simpson: Yes I understand that but comming from gentoo I expected the master to be fine :)
<Guest73314>
Hello, I want to enable some Configuration flags that the Nginx has them at the build for example ( ++ optional withStream "--with-stream")
<clever>
but still no evals finishing
<simpson>
S0rin: Also, if you learn how to use git, more opportunities open up. I use a git branch which very loosely tracks master in order to manage my cloud machines; I can rebase the branch periodically.
<Guest73314>
Any idea how I do this?
<clever>
Guest73314: create a packageOverride that does nginx = pkgs.nginx.override { withStream = true; };
<Guest73314>
awesome! thanks
<simpson>
S0rin: As somebody who used to Gentoo quite a bit, nixpkgs is roughly as stable IME. Much more stable when it comes to actual compilation failures. But YMMV; you may use different packages than me, and nixpkgs is big.
simukis has joined #nixos
<ottidmes>
S0rin: The basic idea is that you point the the location of the unstable branch and take the packages from there. I have a nixpkgs git checkout for both stable and unstable
<S0rin>
simpson: Yes it is fine but I get some package that regulary come back broken (mailutils), and it should be better to only upgrade the needed one
<S0rin>
ottidmes: oh right does it can fit into the configuration.nix ?
<ottidmes>
Right now I have: (import /my/dev/proj/nixpkgs-unstable { }).xdo for example, but I that was a quick and dirty solution, I normally have in my nixpkgs config in packageOverrides an entry for unstable, which is just defined like unstable = import /my/dev/proj/nixpkgs-unstable { };
zeus_ has joined #nixos
newhoggy has joined #nixos
<clever>
gchristensen: bit of a mouthful now, but it looks like the older method will still work, if you dont care about the hash
<S0rin>
ottidmes: oh I see
Neo-- has quit [(Remote host closed the connection)]
<simpson>
S0rin: It does sound like recent unluck; there's only been one actual commit to master about this, so the rest of it is channel fun.
<S0rin>
ottidmes: ok I will try to do that thank you !
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<S0rin>
simpson: yes I am often unlucky :/
zeus_ has quit [(Ping timeout: 246 seconds)]
<Guest73314>
@clever, thanks, it worked like a charm
newhoggy has quit [(Ping timeout: 248 seconds)]
Guest73314 has quit [(Quit: Page closed)]
<S0rin>
But I am trying to learn solutions I can apply in such case (overlay, multi-channels, version pinning) or other if available
<ottidmes>
S0rin: My preferred solution is to have that unstable entry in packageOverrides of the nixpkgs config.nix as I mentioned, and then setup my configuration.nix to have nixpkgs.config import my config.nix so they are the same. Then you can use unstable.package-name in both your nixpkgs config.nix and in configuration.nix via pkgs.unstable.package-name
<S0rin>
ottidmes: That seems fine like a more simple way to pin the version of an other overlay
<simpson>
S0rin: Pinning isn't really a thing, because most packages in nixpkgs don't have multiple versions. Overlays are a thing but I don't use them and I don't know how well they work. I only hear about the Mozilla overlay, really. Multi-channel can be a pain because of overlap.
<simpson>
S0rin: My main strategy is nix-shell everywhere, nearly nothing in nix-env, and always work from a local nixpkgs checkout.
<ottidmes>
Exactly that is what I would recommend as well
newhoggy has joined #nixos
Itkovian has joined #nixos
ryanartecona has quit [(Quit: ryanartecona)]
<ottidmes>
And with nixos stable and unstable in local checkouts, you have full control when to update, I have yet to had any problems with anything broken since I have been using that approach
<S0rin>
simpson, ottidmes Ok I am slowly accepting that, I think using only different overlay can be fine I have to test to be sure
stanibanani has joined #nixos
bennofs has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
michaelpj has quit [(Read error: Connection reset by peer)]
<S0rin>
ottidmes: it is basically what I did in gentoo, but I had version pinning to help in case where the overlay sudenly change
therp has joined #nixos
michaelpj has joined #nixos
<therp>
Hi community, does somebody know why the nixos-unstable channel is broken for >10 days?
<S0rin>
I really like nixos, it is not negative critic, I think the design is more modern and more solid, but when I compare with gentoo in terms of feature the main thing I regret is the local overlay approch where you can overide anything
Turion has quit [(Read error: Connection reset by peer)]
jbgi has joined #nixos
<clever>
bennofs: i see an eval from 3 days ago that fully passed, so it should have updated
<bennofs>
S0rin: hve you looked at overlays?
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<bennofs>
clever: oh right...
elninja44 has joined #nixos
<S0rin>
bennofs: Yes but I did not figure it out to solve my problem maybe I am slow
<bennofs>
S0rin: I generally use nixos-unstable and cherry-pick whatever I need from master
newhoggy has joined #nixos
<bennofs>
S0rin: theoretically, it should also be possible to do something like: let pkgsStable = import <nixpkgs> {}; pkgsUnstable = import <nixpkgs-unstable> {}; in ... if you setup your NIX_PATH correctly
<bennofs>
S0rin: then you can do systemPackages = [ pkgsStable.foo pkgsUnstable.mailutils ]
<S0rin>
bennofs: If I install a package only in unstable and not in stable it should not be problematic to the nix store right ?
<bennofs>
S0rin: what do you mean by problematic to the nix store?
<S0rin>
bennofs: creating a lot of duplication
<bennofs>
S0rin: hmm, you'll probably have a lot of duplication, that's right. But that's just because unstable packages depend on unstable libraries as well, and unstable libraries have probably changed at some point
<ottidmes>
S0rin: No, it should not, they are just descriptions of packages, does not matter what the source is (your own, stable, unstable, master), and if the description is different from another, it will have an unique hash
newhoggy has quit [(Ping timeout: 240 seconds)]
Ivanych has joined #nixos
<simpson>
S0rin: Part of the store's purpose is to accumulate garbage.
<S0rin>
bennofs, simpson Thanks I just have to solve the way to write it down in my configutation.nix to test
<bennofs>
For example, i'm reasonable sure that between unstable and stable, glibc will have a different hash so everything will have a different hash so almost nothing can be shared
<S0rin>
bennofs: Yes but if that work I can have a multi-channel to override
<ottidmes>
I have been mixing stable and unstable for a few years, never had a problem other than making sure you do not have too many deps on unstable, otherwise you end up with a too much duplication. In that case I always modified my local checkout of the stable branch to incorporate those things from unstable, so they are once again shared
<S0rin>
do you think it is a lot of work to package btrfsmaintenances script for nixos ?
<S0rin>
ottidmes: You can do it once in a while but you have to check btrfs-scrub to be sure your drive is not actually dying
proteusguy has quit [(Remote host closed the connection)]
newhoggy has quit [(Ping timeout: 240 seconds)]
<ottidmes>
S0rin: Regarding disk space, in general NixOS will use a lot more than the average OS. I always have to garbage collect when I have a big update, otherwise I run out the limited space on my SSD
<pbogdan>
is there any documentation on how to build a kernel that includes custom patches on NixOS?
<S0rin>
ottidmes: Yes but nixos can rollback easily so you cant have anything
newhoggy has joined #nixos
<ottidmes>
S0rin: Indeed, well worth its price
proteusguy has joined #nixos
<S0rin>
ottidmes: but as /nix is read-only i think you should activate the fstab switch to be able to autodefrag anyway it is mostly usefull when filling the disk
<S0rin>
(because you can't defrag manually the store)
<ottidmes>
S0rin: I started a scrub, I am curious what will come out of it, but my system is basically on its last breath anyway (8 years old)
therp has quit [(Quit: Leaving)]
<S0rin>
ottidmes: scrub can tell you if the hash of a object (file) is different actually than recorded before usually usefull to detect dying hard-disk
newhoggy has quit [(Ping timeout: 240 seconds)]
<S0rin>
ottidmes: I wish you luck had to RMA a ssd after failling the btrfs scrub last month
<S0rin>
this is where raid-1 (even in plain btrfs) worth its price
<ottidmes>
S0rin: 0 errors, yay!
<S0rin>
ottidmes: :)
<ottidmes>
S0rin: I will see if I can package that btrfsmaintenance repo for you
<S0rin>
btrfs is great but still lacking some admin tools to check his health
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aszlig pushed 1 new commit to master: https://git.io/vQFrW
<ottidmes>
S0rin: I e.g. have nix.nixPath = singleton "nixpkgs-unstable=/my/dev/proj/nixpkgs-unstable"; to add it to my NIX_PATH
Itkovian has joined #nixos
<clever>
S0rin: are you checking roots channel list or your own user?
<LnL>
S0rin: you need to add the channel as root
<S0rin>
LnL: I did it as root (with sudo)
<Infinisil>
Oh, fancy, nix-store --optimize just saved me 10GB
<clever>
S0rin: did you also nix-channel --update as root?
<clever>
Infinisil: there is also an option to make nix internally run it after every operation
<S0rin>
clever: no I've just sudo nixos-rebuild switch --upgrade
<LnL>
oh right you might need to run update before it shows up
zraexy has joined #nixos
<clever>
Infinisil: auto-optimise-store = true
lvernschrock has quit [(Quit: Simple IRC: The quit option.)]
<clever>
S0rin: yeah, like LnL said, --add only takes effect after --update
newhoggy has quit [(Ping timeout: 246 seconds)]
<S0rin>
LnL, clever yes I've did it !
bara has quit [(Remote host closed the connection)]
jbgi has quit [(Ping timeout: 246 seconds)]
acarrico has joined #nixos
<S0rin>
It is rebuilding let's see if it works out
<S0rin>
I've finally used a stable base with some package from unstable
necronian1 has joined #nixos
DutchWolfie has quit [(Quit: Konversation terminated!)]
<Infinisil>
Umm, how do I modify /etc/nix/nix.conf?
<clever>
Infinisil: nix.extraOptions in configuration.nix
<Infinisil>
Ahh
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
newhoggy has joined #nixos
<Infinisil>
Oh, nix.autoOptimizeStore = true
slyfox_ is now known as slyfox
<clever>
oh, that sounds better
<clever>
i was using my router as an example, and its in dire need of an update
jbgi has joined #nixos
<ottidmes>
I have been using that setting as well, but just optimized 2G
necronian has quit [(Quit: WeeChat 1.9)]
olejorgenb[m] has joined #nixos
Mercuria1Alchemi has quit [(Ping timeout: 246 seconds)]
bara has joined #nixos
Itkovian has joined #nixos
necronian1 is now known as necronian
newhoggy has quit [(Ping timeout: 260 seconds)]
bhipple has quit [(Quit: WeeChat 1.7.1)]
Itkovian has quit [(Client Quit)]
jsgrant has joined #nixos
olejorgenb[m] has left #nixos ["User left"]
newhoggy has joined #nixos
ibrahims has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
Nobabs27 has joined #nixos
<ottidmes>
What is preferred when packaging, using install or mkdir/cp?
<gchristensen>
ottidmes: no preference
<ottidmes>
gchristensen: Alright, then I will stick with mkdir/cp since I am used to doing so
<gchristensen>
cool
<gchristensen>
install definitely hass an advantage of cp in terms of permission settingg of course
newhoggy has joined #nixos
<ottidmes>
gchristensen: I was wondering about that, if it were better, why not promote its use? Arch Linux seems to consistently use install
ixxie has quit [(Quit: Lost terminal)]
Nobabs27 has quit [(Quit: Leaving)]
<gchristensen>
good and fair question :)
<bennofs>
well except for the executable bit, nix strips permissions anyway
<gchristensen>
ohyeah
newhoggy has quit [(Ping timeout: 248 seconds)]
newhoggy has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] teh opened pull request #27432: pytorch: init at 0.1.12 (master...pytorch) https://git.io/vQF66
NixOS_GitHub has left #nixos []
zeus_ has joined #nixos
<ottidmes>
bennofs and gchristensen: Thanks for clearing that up, now I feel safe using mkdir and cp without regard of permissions
newhoggy has quit [(Ping timeout: 240 seconds)]
Itkovian has joined #nixos
oida has joined #nixos
<ottidmes>
What is the best way to replace multiple lines from a script during the patch phase, I could use a --replace with substituteInPlace for each of the lines, other than an actual patch, are the other recommended ways?
<gchristensen>
what sort of lines?
Itkovian has quit [(Client Quit)]
<ottidmes>
umask 022\nPATH=...\nexport PATH
newhoggy has joined #nixos
zeus_ has quit [(Ping timeout: 246 seconds)]
oida_ has quit [(Ping timeout: 260 seconds)]
newhoggy has quit [(Ping timeout: 268 seconds)]
<ottidmes>
Why is buildInputs called buildInputs? I mean the very first sentence describing it is "A list of dependencies used by the new derivation at run-time.". Why then not call it runtimeDeps?
<ottidmes>
Or runtimeInputs
<bennofs>
ottidmes: ? it's not really runtime deps either. Nix scans the output afterwards to determine runtime dependencies
newhoggy has joined #nixos
<ottidmes>
bennofs: So if I package some script, do I still need to make sure I add e.g. findutils for the find command or is it able to determine those?
<bennofs>
ottidmes: even runtime dependencies are not added to the user environment
<srhb>
ottidmes: It is supposed to make you think of them as function inputs / arguments to the build process.
<srhb>
ottidmes: I think that it's a decent name for that.
<bennofs>
ottidmes: the best way to package an existing script is to wrap it with a wrapper that adds all necessary utils to PATH before executing the script
newhoggy has quit [(Ping timeout: 268 seconds)]
<ottidmes>
srhb: Yeah, ok, that first sentence just threw me off
newhoggy has joined #nixos
<srhb>
ottidmes: Oh, sorry.
<ottidmes>
bennofs: Ah, thanks for the explanation. It is clear to me now, they are truly build inputs, and it will check the resulting binaries for their runtime dependencies and those will be added to the runtime environment and since this cannot be inferred from a script, you will have to do this explicitly, right?
<bennofs>
ottidmes: it won't even add it to the runtime environment really. It will just make sure that the referenced paths exist in the store
<bennofs>
ottidmes: since everything is referenced by absolute paths, there is no need to "add" anything to some search path or similar
<ottidmes>
bennofs: Ah right!
<bennofs>
wow, I didn't know hydra.nixos.org runs GNU emacs CI :)
newhoggy has quit [(Ping timeout: 255 seconds)]
cherrybl0ss0m_ has quit [(Quit: Connection closed for inactivity)]
<ottidmes>
bennofs: The correct way to set the PATH for those already existing scripts is wrapProgram "$file" --set PATH "... the /bin folders of the deps ...", right?
<ottidmes>
Infinisil: Thanks, I was just about to use it
newhoggy has joined #nixos
<bennofs>
can we get back to 300 open PRs?
<bennofs>
:D
Ivanych has quit [(Ping timeout: 240 seconds)]
sciguy has joined #nixos
newhoggy has quit [(Ping timeout: 248 seconds)]
<sciguy>
hi all. sorry to just pop in and start firing questions, but I am looking into utilizing Nix (pkgs) in the context of computational scientific work.
<bennofs>
is cache.nixos.org supposed to support pings?
<gchristensen>
bennofs: yes
<Infinisil>
bennofs: Should I just look at some PR's and test them myself/review them a bit? I'm not privileged in any way, but I think this would help too, right?
<sciguy>
i would like to be able to (as easy as possible) provide a way for users to build my code and be dropped into an environment with all the exact dependencies originally used. can someone point me in the right direction?
<bennofs>
Infinisil: i'm not sure. most of them look to me as if they just need someone to press the merge button TBH
<Infinisil>
bennofs: I think people with merge-right would be more willing to do that when somebody else confirmed the changed
<bennofs>
Is 2600:9000:200d:b600:10:dfa4:b0c0:93a1 down for anyone else?
newhoggy has joined #nixos
<bennofs>
(d3m36hgdyp4koz.cloudfront.net)
<LnL>
works fine here
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb created wireguard-private-keys (+1 new commit): https://git.io/vQFXd
<NixOS_GitHub>
nixpkgs/wireguard-private-keys c3cb467 Aristid Breitkreuz: wireguard: allow not storing private keys in world-readable /nix/store
NixOS_GitHub has left #nixos []
<Infinisil>
sciguy: I'd like to know too about this. Just using import <nixpkgs> wouldn't work as this doesn't fix the versions. I believe importing from a specific revision of nixpkgs from github would work
<bennofs>
(note the builtins.fetchTarball used to fetch a specific nixpkgs revision)
\\\\\\\\\\\ has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
<bennofs>
LnL: did you try the IPV6 address or just the domain?
<bennofs>
looks like my provider messed up some IPv6 then
<LnL>
both
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb opened pull request #27433: wireguard: allow not storing private keys in world-readable /nix/store (master...wireguard-private-keys) https://git.io/vQF18
NixOS_GitHub has left #nixos []
<Infinisil>
bennofs: I like your approach
<sciguy>
thank you for the link. would the rest of the process then be more or less like a normal default.nix file? In that i should be able to just fetch my code from github, and with the pinned nixpkg version, be assured they grab the same dependencies?
<bennofs>
alright, I cannot ping google.de either. so my provider messed up
<sciguy>
sorry, iv only been reading about nix for the past 2-3 days
Itkovian has joined #nixos
<Infinisil>
sciguy: If you want to develop locally you probably want to use `src = ./.`
<ottidmes>
I like to use nix-shell -E '(import <nixpkgs> { }).callPackage ./. { }' to work on a certain default.nix without modifying the default.nix file, you could modify the command to include a call to fetchTarBall as bennofs showed instead of <nixpkgs>
<bennofs>
i like to write my default.nix files for packages outside of nixpkgs in a way that just nix-shell --pure will work
<sciguy>
my main intention is to be able to say "look, click here", and a researcher will be dropped into an environment in which he/she can re-run my experiments in a reproducible manner
newhoggy has joined #nixos
<Infinisil>
sciguy: Do you assume a nix installation?
<sciguy>
i think that owuld be a reasonable assumption
<ottidmes>
bennofs: But don't you then end up with two default.nix files, one modified to work with nix-shell and one for nixpkgs?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] grahamc pushed 2 new commits to master: https://git.io/vQF1S
<NixOS_GitHub>
nixpkgs/master c2b0cb5 Benjamin Staffin: nixos: nix snowflake logo for the nixos manual launcher
<NixOS_GitHub>
nixpkgs/master 6b879ef Graham Christensen: Merge pull request #23964 from benley/nixos-manual-launcher...
NixOS_GitHub has left #nixos []
<Infinisil>
sciguy: Do you expect them to modify the experiments or read-only?
<sciguy>
ideally i would like the possibility for both. but my first priority would be read-only
<bennofs>
ottidmes: yes. if the project is already in nixpkgs, I often put only a `with (import <nixpkgs> {}); myPackage.overrideAttrs { ... }` in the project's default.nix
zraexy has quit [(Ping timeout: 268 seconds)]
<bennofs>
sciguy: writing a default.nix for your experiment setup is probably the best way here. then users can just nix-shell --pure and they'll get a bash with everything setup as you wish
<Infinisil>
Yup, and put this in a git repo, so they can clone it locally, then nix-shell and ready to go
c74d has quit [(Quit: c74d)]
<sciguy>
sounds relatively painless once i get the syntax and workings of NIX down
newhoggy has quit [(Ping timeout: 240 seconds)]
<sciguy>
thanks
<Infinisil>
You could also just distribute a single nix file that fetches the source itself, but then it's read-only
newhoggy has joined #nixos
c74d has joined #nixos
<sciguy>
is anyone awake if there have been any publications or work done in the direction of using NIX for reproducible computational experiments?
<sciguy>
I have seen plenty for general development environments, but not in so much in the direction of scientific work
<ottidmes>
sciguy: It is quite annoying when you find some cool paper with an implementation only to find out that is impossible to get it working on your system. But you still end up with sources of dependencies no longer being available eventually, dont you?
<Infinisil>
ottidmes: If every dependency is using git and doesn't randomly change hashes, then I think there shouldn't be a problem
<ottidmes>
Infinisil: And those git repos remain available you mean
katyucha_ has joined #nixos
katyucha is now known as Guest6867
Guest6867 has quit [(Killed (wilhelm.freenode.net (Nickname regained by services)))]
katyucha_ is now known as katyucha
katyucha_ has joined #nixos
<Infinisil>
ottidmes: Yes
bennofs has quit [(Ping timeout: 240 seconds)]
<Infinisil>
It would be possible to copy the closure of the dependencies, then it will work even more certainly
<ottidmes>
It is the older implementation that generally are impossible to get working, and if years pass, some of those git repos might be removed, so you would need to package the sources as well if you really want to have a solid solution
darlan has quit [(Client Quit)]
<sciguy>
best case we could package all dependencies in our own private repo and use NIX to grab and build
<sciguy>
im thinking thats possible somehow with NIX
<NixOS_GitHub>
nixpkgs/master 6504df6 Graham Christensen: Merge pull request #25955 from matthewbauer/nixpkgs-manual-declarative-package-management...
NixOS_GitHub has left #nixos []
<ottidmes>
sciguy: In the past I have setup my desktop to be a binary cache for my server, since they had a lot of overlap and I did not want to build on my server if it could be helped
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<ottidmes>
Infinisil: You need to add a binary cache as well, so if you host your own binary cache, you can be certain people will be able to download the stuff
<gchristensen>
bennofs[m]: ok I merged some :P
newhoggy has quit [(Ping timeout: 240 seconds)]
<bennofs[m]>
!m gchristensen
<[0__0]>
You're doing good work, gchristensen!
<Infinisil>
ottidmes: So only the source, a channel and a binary cache :P. I think it would be easier to distribute the closure
<boomshroom>
I'm trying to build a derivation including some pending patches to LLVM, but when it trys to apply them, it complains that some of the files are missing, where as unpacking the archive manually, I can find them easily.
Jackneill has quit [(Remote host closed the connection)]
<Infinisil>
boomshroom: Link/gist?
<gchristensen>
Infinisil: will people not usually want the gui?
erictapen has joined #nixos
<Infinisil>
gchristensen: I'm not sure, I thought maybe the nix people would prefer the cli. The only thing the cli doesn't provide easily is visualization of diagrams. The gui is very very basic, doesn't even have syntax highlighting
<Infinisil>
gchristensen: Well actually I might as well include it by default, it's not like it has many dependencies anyways and the source is the same too
<gchristensen>
:)
<Infinisil>
And the binaries are of a different name, scyther-cli and scyther-gui
<boomshroom>
I run patch manually and it appears to hang.
<gchristensen>
Infinisil: new question: is there any good reason a user might want to turn off the gui support? it addds quite a lot of complexity to the file
erictapen has quit [(Ping timeout: 260 seconds)]
<boomshroom>
Minor curiosity: is it currently possible to build a NixOS system from within a system with Nix?
<clever>
boomshroom: as a vm, or as just a dir in the store?, or a disk image?
newhoggy has quit [(Ping timeout: 260 seconds)]
<Infinisil>
gchristensen: Minimal installation kind of thing, the gui doesn't add anything significant and is completely useless without X
<gchristensen>
aye
<gchristensen>
just trying to assess if that use case is worth the complexity
<gchristensen>
but it isn't so bad
<boomshroom>
clever: onto a new partition. (It would probably have to be pre-formatted and mounted)
<Infinisil>
gchristensen: Valid thought, the package won't need to be updated anyways though, last release was in like 2014
<gchristensen>
:)
<clever>
boomshroom: ah, then you want the linode guide
<clever>
boomshroom: let me check my irc logs
<boomshroom>
I guess it would basically just be like the LiveCD/DVD/USB, but within an already installed *nix system. I don't care much about it, just something that idly crossed my mind.
michaelpj has quit [(Read error: Connection reset by peer)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] aristidb pushed 1 new commit to wireguard-private-keys: https://git.io/vQFye
<dtzWill>
basically I was looking for patches on master but not release_40 in the lib/Target/RISCV directory
<Infinisil>
This bot should post the open PR count as well :D
stanibanani has quit [(Ping timeout: 268 seconds)]
<boomshroom>
dtzWill: LLVM is useful, but it's so complex that it's difficult to add a new backend. Does anyone know if there exists a simple retargetable compiler? I don't care about optimizations, because those add complexity and right know I just want to get something running.
<pstn>
Neato. New tt-rss.
<Infinisil>
boomshroom: Do you mean backend as in a new architecture?
<dtzWill>
boomshroom: do the RISCV guys have a fork/branch somewhere with the patches applied? Seems like something they probably have...
<Infinisil>
Wizek__: You don't have write access to the file?
<Infinisil>
Wizek__: Or if you need more complicated rules, then I suggest you to use a proper default.nix or shell.nix file
<dtzWill>
bbiab
<Wizek__>
Infinisil: I do. But often I need to run it with one executable, and occasionally another. So I am hoping for being able to leave a defaut and also be able to override
hiratara has quit [(Ping timeout: 246 seconds)]
<boomshroom>
I can't imagine building 2 different LLVM installations simultaneously is very healthy for my computer.
hiratara has joined #nixos
newhoggy has joined #nixos
<clever>
Wizek__: yes
<Infinisil>
Wizek__: Maybe something like `nix-shell -p ghc --run 'runhaskell ./test.hs'` could work for you
<Wizek__>
clever: I believe -i and -I do distinct things, don't they?
<clever>
yeah, thought you meant the uppercase one
<boomshroom>
I'm trying to build RedoxOS on nix and I'm thinking of making a nix derivation. One thing about it is that it doesn't have an install (for good reason), but rather has a make qemu. What would be the appropriate why to handle this? Should the install phase simply copy over the entire source directory?
<Infinisil>
boomshroom: Shouldn't you put the compiled files in $out?
ng0 has quit [(Quit: Alexa, when is the end of world?)]
<boomshroom>
Infinisil: The thing is that running it uses make.
<Infinisil>
boomshroom: Hmm, I'll have a closer look
thc202 has quit [(Ping timeout: 248 seconds)]
<Infinisil>
boomshroom: There seems to be a `make iso`, wouldn't that be more fitting as a build?
<Infinisil>
boomshroom: And nixops could be used to deploy to a VM
<Infinisil>
boomshroom: No idea, but this seems pretty flexible: ttps://github.com/NixOS/nixpkgs/blob/master/pkgs/top-level/all-packages.nix#L6010-L6023
katyucha has quit [(Remote host closed the connection)]
yurrriq has joined #nixos
philips has joined #nixos
nand0p has joined #nixos
eddyb has joined #nixos
luto has joined #nixos
leat has quit [(Ping timeout: 240 seconds)]
proteusguy has quit [(Ping timeout: 258 seconds)]
schoppenhauer has joined #nixos
<schoppenhauer>
hi. is there a lxc template for nixos?
k0001 has joined #nixos
<gchristensen>
schoppenhauer: there might be something in the issue tracker
<NixOS_GitHub>
[nixpkgs] apeyroux opened pull request #27437: ipfs: fix #27414 - fusermount": executable file not found (master...fix-ipfs-fusermount) https://git.io/vQFQy
NixOS_GitHub has left #nixos []
hiratara has joined #nixos
newhoggy has quit [(Remote host closed the connection)]
<schoppenhauer>
gchristensen: ok
<schoppenhauer>
thx
lambdael has joined #nixos
<timclassic>
Can someone point me in the right direction for getting the ZFS kernel modules loaded in the live installer?
<gchristensen>
sure, timclassic
<Infinisil>
timclassic: I think adding boot.supportedFilesystems = [ "zfs" ]; in /etc/nixos/configuration.nix then rebuilding should do, no?
<gchristensen>
^
newhoggy has joined #nixos
<gchristensen>
and to be extra clear, /etc/nixos/configuration.nix, being, intentionally, the one in the installer, not the one in/mnt
<Infinisil>
Exactly
<timclassic>
Oh interesting, okay
<Infinisil>
How can I make a rebuild fail if a specific systemd unit that changed failed to start? For example it happens with udev-settly a lot lately
<timclassic>
This is obviously my first direct install :)
<gchristensen>
timclassic: neat! let us know if you have any problems :)
<gchristensen>
timclassic: are you putting / on zfs?
<timclassic>
Infinisil: gchristensen: I'll give it a go, thanks.
<timclassic>
/ is the plan, yes
<timclassic>
Be back soon...
<gchristensen>
cool
<gchristensen>
timclassic: oh wait
newhoggy has quit [(Remote host closed the connection)]
<gchristensen>
I have to leave here in 10min but I have some helpful scripts for you for reference
<Infinisil>
gchristensen: I'd like to have a look at these too, gonna redo my filesystem soon
<gchristensen>
this is purely anecdotal, out of date, and biased, but personally I've never lost data due to zfs, and have lost dozens-to-hundreds of servers due to btrfs corruption
<Infinisil>
gchristensen: That is a pretty good reason
<gchristensen>
I haven't tried btrfs for a few years now, though
<ottidmes>
I thought now a days BTRFS is pretty solid, I have been running it for a few years on both my desktop and servers without any loss, but I have not been making a lot use of its more advanced features
<gchristensen>
I found it a very effective chaos monkey -- good at taking down prod servers.
<boomshroom>
Another reason is that ZFS is supported on Mac OSX.
<Infinisil>
gchristensen: Are you using ZFS on the full disk then?
<ottidmes>
For my backup server I probably will go for ZFS, just in case, considering its purpose and all those horror stories about BTRFS, but my other machines I probably will stick with BTRFS
<gchristensen>
Infinisil: check out the partitioning -- I create one big partition and use that forzfs
<Infinisil>
gchristensen: Pretty cool, I might do that too, I couldn't until now because I had to share the disk with macOS
<Infinisil>
I just bope I don't screw up booting
<gchristensen>
I'm not sure you specifically need to make the one partition first, I could try not doing that
<gchristensen>
just ran out of patience :P
<boomshroom>
One LP episode later, both atempts at building LLVM with RISC-V patches succeeded and then segfaulted when tried to compile a simple ir file to assembly.
<Infinisil>
gchristensen: Probably just doing `zpool create -f sda` would work too
<gchristensen>
yeah
<Infinisil>
gchristensen: But the bootable bit might be important
<gchristensen>
it is just a bit of no fun testing that on some of these instance types, especially the type 2 which takes about 5-10 minutes to initialize the disks on boot
<gchristensen>
so when it worked I stopped poking :P
jgertm has quit [(Ping timeout: 268 seconds)]
* benley
<--- uses zfs root on laptops
<benley>
it's pretty good.
<Infinisil>
benley: Nice, using UEFI booting and grub?
<benley>
yup. And LUKS disk encryption beneath zfs.
* c74d
uses ZFS root on desktops and laptops, using UEFI and systemd-boot
<benley>
I couldn't figure out how to get systemd-boot working with /boot on encrypted ZFS, so I'm using grub
<Infinisil>
benley: Oh very nice to hear, any tips on getting it set up correctly?
<benley>
"don't fuck up"
<Infinisil>
Heh
<avn>
Infinisil: it works using uefi and grub, but grub unable to read kernels from /nix/store on ZFS (if you use hardlinks)
<benley>
I'm not sure if encrypted /boot is actually worthwhile unless you can also get some form of Secure Boot working to prevent bootloader tampering
<Infinisil>
avn: Aren't they installed to /boot using softlinks?
<Infinisil>
the kernels
<benley>
I'm pretty sure they are just straight up copied into /boot/kernels
<avn>
Infinisil: depend from copyKernels value
<benley>
along with the initrd images
<Infinisil>
benley: Wait so you have zfs at the bottom and then luks?
<benley>
Oh, I didn't know about copyKernels
<avn>
It can read right from store (in theory), but really not -- due bug in grub' zfs
<benley>
no, I have luks at the bottom, then ZFS on top of luks
<benley>
someday ZFS will have its own encryption, but that isn't ready for use yet
<benley>
zram swap is neat, but I wish there was a way to avoid the overhead of having to evict pages from main memory, compress them, then write them back to "swap" which is also in ram
<benley>
i.e. just compress them from the get go and amortize the performance hit
<benley>
(maybe that hit is worse than I imagine it would be? dunno)
<Infinisil>
benley: You mean compressing the whole ram all the time?
<benley>
yeah
<Infinisil>
That would be horrible for performance
<benley>
I really don't know if that would be a good idea or not. Probably not unless you were really short on memory.
<benley>
if you were always digging into swap, always compressing might end up being faster
<Infinisil>
Hmm maybe
<Infinisil>
I may just use a zvol for a swap
<Infinisil>
I think I need a non-ram swap for things such as hibernation
<benley>
oh hm, I wonder if that actually works
<benley>
right now if I try to hibernate my laptop just falls over and dies
<benley>
(turns out hibernating to zram doesn't work so well)
nixos-users-wiki has quit [(Remote host closed the connection)]
<timclassic>
gchristensen: Thanks!
nixos-users-wiki has joined #nixos
<Infinisil>
The thing is that hibernation should work even after power outage, because it wrote it to disk
<Infinisil>
Oh wait you are using an disk swap
<Infinisil>
No idea why it fails then
Wizek_ has quit [(Ping timeout: 246 seconds)]
jdz1200 has joined #nixos
<avn>
Infinisil: better not use zvol swap, I tried -- and it followed to unpredictable behavior on memory pressure
ibrahims has quit [(Ping timeout: 240 seconds)]
<jdz1200>
Are failures like these expected (known bug) when using NixOps on MacOS and deploying to VirtualBox? https://pastebin.com/raw/CmTPtC1f
joehh has quit [(Ping timeout: 260 seconds)]
<jdz1200>
WIll ask in nix-darwin, looking at the banner
<Infinisil>
avn: Hmmm.. So either no swap or no full zfs disk... :(
<c74d>
I choose swap; what are the benefits of a "full ZFS disk"?
<Infinisil>
c74d: I don't know the internals, there are some optimizations
<Infinisil>
I think it's in one of the man pages
<avn>
Infinisil: I just added smaller disk for swap, and stuff which need random write (like torrents)
<Infinisil>
avn: I'm using a laptop so I can't just use a separate disk for swap
<c74d>
Infinisil: the one I remember is that then ZFS can set the I/O scheduler for the disk to `noop`, preventing Linux's I/O scheduling from interfering with ZFS's own I/O scheduling — but one can set that manually easily enough
<Infinisil>
c74d: Ah here, quoting `man zpool`: ZFS can use individual partitions, though the recommended mode of opera‐
<Infinisil>
tion is to use whole disks.
markus1189 has joined #nixos
<Infinisil>
c74d: I definitely want to try to be as efficient as possible with my old laptop (but also safe)
boj has joined #nixos
<avn>
Infinisil: I have 2 disks with 2 parts each efi+zpool
markus1199 has quit [(Ping timeout: 240 seconds)]
<Infinisil>
jdz1200: That's actually the same thing that's happening for one of my PRs:
<Infinisil>
i also have no idea why that's happening
<jdz1200>
Ah. Interesting
<Infinisil>
jdz1200: I set platforms to linux only, yet travis tries to build it on osx
<Infinisil>
Oh well actually I think it's a completely different problem, just the same error message
<jdz1200>
Agreed. Thanks for the link though
zeus_ has joined #nixos
<Infinisil>
gchristensen: I'm noticing that you're using `zpool create ... /dev/sdc1` so I have the suspecion that you're not actually using the full zfs disk optimizations
<Infinisil>
(Searched for bcache before, that's why i didn't see anything)
takle has quit [(Remote host closed the connection)]
Wizek_ has joined #nixos
zeus_ has quit [(Ping timeout: 246 seconds)]
<boomshroom>
One of the things I love about NixOS is how customizable it is using such a simple syntax.
<Infinisil>
boomshroom: Indeed, and if something's missing you can mostly do it yourself without much effort
<boomshroom>
Infinisil: I more so mean swapping in and out large portions of the system in just a few lines.
<Infinisil>
Ah yes
<ottidmes>
I do not dare to remove my Arch Linux installation on my old laptop, because I no longer know exactly what changes I made to it along the years, with Nix this is explicit, and you get a history if you use source control