<aleph->
Hey building a go package with buildGoPackage. Would I build it like: nix-build -E 'with import <nixpkgs> ().go.CallPackage ./default.nix {};'
<ottidmes>
aleph-: you would call it just like normal AFAIK
<noonien>
i'm trying to cross-compile a binary for a target that does not run nixos
<noonien>
it's a go binary, and for some reason `-dynamic-linker /nix/store/jdqssyw02jpvyvhgjks4xlv510zz5hrn-glibc-2.27-armv7l-unknown-linux-gnueabihf/lib/ld-linux-armhf.so.3 ` gets passed as an argument to gcc i believe
<noonien>
this results in the binary not running on the target system, because there is no nix store
<ottidmes>
how do I undo the --insecure flag passed to curl, there doesnt seem to be a --secure/--no-insecure
<noonien>
ottidmes: you can't as far as i can tell
<{^_^}>
[nixpkgs] @tek opened pull request #55922 → fix syntax error in tt-rss pre-start script → https://git.io/fh5z1
<clever>
noonien: you want static linking then
<noonien>
however, i believe all files that get downloaded via the network are verified by hash
<clever>
noonien: pkgsCross.muslpi
<noonien>
clever: hmm, not really, i just want the correct path for ld-linux.armhf.so.3, which should be in /lib/
<clever>
ottidmes: yeah, nix verified the hash of all fixed-output things, so a mitm attack can only (waste bandwidth|see what your doing|block downloading it)
<clever>
noonien: then use patchelf to change it after the build, almost exactly the same as when your packing stuff for nixos
<noonien>
i see
<ottidmes>
clever: I know, but when applying TOFU you actually want it enalbed
<noonien>
what triggers this behavior? the way gcc was built?
<clever>
ottidmes: ah, nix-prefetch-url then
<clever>
noonien: the cc-wrapper script
<aleph->
Okay so my package isn't seeing the internal files it's trying to reference.
<aleph->
Hmm.
<ottidmes>
clever: for example, but nix-prefetch-url is insecure too, right?
<clever>
ottidmes: i think that one is secure
eadwu has joined #nixos
justanotheruser has joined #nixos
<ottidmes>
clever: yep, you are right, I checked the source and seems to be the case, builtins.fetchurl is insecure (makes sense considering your previous argument). I'll keep it as a plan B
<clever>
ottidmes: builtins.fetchurl doesnt use the curl command, but rather, the curl library
<{^_^}>
[nixpkgs] @jmgrosen opened pull request #55923 → spotifyd: init at 0.2.4 → https://git.io/fh5zS
<ottidmes>
clever: why? its checks by default, but disables it for the builtin fetchurl, as all fetchers have it disabled by default, you could argue though that is should only be disabled when an actual hash has been given
o1lo01ol1o has quit [Ping timeout: 245 seconds]
<clever>
ottidmes: i think its because that fetcher is ran within the normal build sandbox, and can lack CA certs
<clever>
and yeah, it should really only be disabling if a sha256 is present
drgbr has quit [Client Quit]
<ottidmes>
clever: I am asking about this, because its one of the last features I want to implement for my nix-prefetch project
<clever>
ottidmes: it may help you more to know how the hashing is done, and just hash without the store
<clever>
outputHashMode can either be flat or recursive
<clever>
flat requires that $out be a plain file, and then outputHash is just the raw hash of the file (sha256sum, and your done)
<ottidmes>
clever: yeah, I am well aware of those details after doing this :P though not about the actual implementation
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<clever>
recursive allows it to be a file, directory, or symlink, so it is instead the hash of the nar
<clever>
nix-store --dump $out | sha256sum, and your done
<clever>
the actual name/path of $out doesnt matter, so you can also nix-store --dump /tmp/foo | sha256sum
<ottidmes>
clever: yeah, good points, guess thats better than the standard, lets parse an hash mismatch error approach
<clever>
you can also `nix-store --add-fixed sha256 path` optionally with --recursive, to add a file to the store, and get its hash
gagbo has quit [Ping timeout: 246 seconds]
<clever>
ottidmes: the nix-hash command can also do the same hashing, and convert between base32 and base64
<ottidmes>
clever: for the builtin fetchers I would still have to rely on the error messages AFAIK, so I have added it to my TODO to take a look at, but I think the curl issue is higher prio
<ottidmes>
clever: yep, I used it at some point in my project, not anymore though
<clever>
which builtin fetcher?
<clever>
builtins.fetchurl should be the flat hash, builtins.fetchTarball is the recursive hash
<ottidmes>
well any really, the moment I call builtins.fetchurl for example, it will do all things at once
drgbr has quit [Ping timeout: 258 seconds]
nD5Xjz has quit [Ping timeout: 268 seconds]
thc202 has quit [Ping timeout: 245 seconds]
<ottidmes>
but its only when checking the hash, I know that those builtins dont require an actual hash to work, so should never fail with a hash mismatch normally
<clever>
the builtin fetchers are also serial, rather then parallel, so it can help performance to avoid them
<ottidmes>
I just support them, I try not to use them myself anyways
<ma9e>
what do i do if my build step depends on a certain locale?
o1lo01ol1o has joined #nixos
<clever>
ma9e: just set the locale env vars, and add glibcLocales to the nativeBuildInputs
init_6 has joined #nixos
Ariakenom has quit [Remote host closed the connection]
Ariakenom has joined #nixos
<ottidmes>
clever: its possible :D --no-insecure simply works, had not expected that
Ariakenom has quit [Remote host closed the connection]
<clever>
heh, so simple!
maurer has joined #nixos
<ottidmes>
yeah, was about to ask on #curl, when I thought to myself, always first try before asking, so I just tried passing curlOpts ' --no-insecure -vvvv ' and it now fails on not being able to check for a HTTPS URL, so seems to work, now to figure out how to make it not fail (probably setting the cacert path somewhere)
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fh5gO
lawlesseel has joined #nixos
<ottidmes>
hmm setting --cacert still makes it fail to download, curl: (77) error setting certificate verify locations: CAfile: /nix/store/fna8lggi520pgvdwi8p1xyzcj9706j04-nss-cacert-3.40.1/etc/ssl/certs/ca-bundle.crt (which does exist)
<ma9e>
clever: is it normal to see "cannot change locale" when i run nix-shell? the variable is set correctly after that
drakonis has joined #nixos
<clever>
ma9e: yeah, thats normal
drgbr has joined #nixos
<ottidmes>
hmm, running the curl command done by fetchurl in my shell works, but it fails when done in the script
jluttine has quit [Ping timeout: 255 seconds]
andersk has quit [Remote host closed the connection]
andersk has joined #nixos
goibhniu has quit [Ping timeout: 246 seconds]
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fh5gB
goibhniu has joined #nixos
shibboleth has joined #nixos
iyzsong has joined #nixos
<ottidmes>
clever: found the problem, but no clue how to fix it, in the sandbox, my /nix/store/.../etc/ssl/certs/ca-bundle.crt path does not exist, outside it does, any idea on how to fix that? make some Nix option I could use?
<sb0>
how do I completely erase all postgresql data from a nixos system?
<sb0>
ny hydra install became broken (builds are stuck in "scheduled to be built" phase no matter what I do) and I'm running out of patience. i'm just going to wipe it and reinstall everything.
<Ralith>
nuke /var/db/postgresql, I think?
<clever>
ottidmes: yeah, sandbox-paths
<clever>
sb0: its likely that your build slaves just lack features, no need to wipe the db
<clever>
sb0: hydra is bad at reporting missing features
<sb0>
rm -rf /var/lib/postgresql?
<sb0>
I didn't touch the build slaves and I'm just using the default...
<clever>
sb0: nixpkgs added more required features, and wont run on older build slaves
<sb0>
well I need to reinit the DB anyway because I made a lot of mess there trying to find out WTH was wrong with hydra
<clever>
sb0: there is a hydra-init.service in systemd
<sb0>
yep, but my DB is currently corrupted and hydra-init won't work either
<sb0>
so I just want to wipe all pgsql data and start over
<clever>
ah, then stop postgresql.service, and rm that too
<sb0>
rm /var/lib/postgresql?
winem_ has quit [Ping timeout: 268 seconds]
<clever>
yep
<sb0>
ok thanks
<ottidmes>
clever: I am using in my nix eval call that calls pkgs.fetchurl, --option extra-sandbox-paths /etc/ssl/certs/ca-bundle.crt=/nix/store/fna8lggi520pgvdwi8p1xyzcj9706j04-nss-cacert-3.40.1/etc/ssl/certs/ca-bundle.crt, but when I check for /etc/ssl/certs/ca-bundle.crt the file existance in the fetchurl builder.sh, it does not exist, is what I do not allowed like that?
<clever>
` --option extra-sandbox-paths` only works if you are a trusted user
<ottidmes>
hmm, which is not something I can assume, since it supposed to be a general tool
<ottidmes>
guess I will have to override fetchurl to contain cacert, no?
<clever>
ottidmes: only the fetch's inside nixpkgs can be overriden, the builtin ones cant
<ottidmes>
I am well aware, I am talking about the one in nixpkgs
<sb0>
clever: I'm using release-18.09 and I can't find the updated buildslave requirements in nixpkgs
<freusque>
hello - is there a way to make extra-sandbox-paths mounted with write rights?
<freusque>
I like to use these, but I get only read access.
<ottidmes>
clever: the builtin ones are by definition primitives, I will work around that by just doing the verification before calling the builtin fetchers
<reallymemorable>
but I dont know how to find the name of my wireless card within NixOS's CLI
<gchristensen>
what does ifconfig show
<init_6>
reallymemorable: dmesg | grep wifi ?
<reallymemorable>
init_6: i jsut did that and it seemed nothing happened
<reallymemorable>
ifconfig shows a lot of stuff that I dont know how to paste here since the NixOS machine isnt connect to the internet yet
<gchristensen>
ifconfig shows a line for me like wlp2s0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 -- and the name is wlp2s0
<reallymemorable>
my first line says
<reallymemorable>
lo: flags=73<UP,LOOPBACK,RUNNING> mtu 65536
<reallymemorable>
then there are more lines that say inet...
<reallymemorable>
etc
<delroth>
anyone generating dns zones with their nixos configuration? if so, how do you handle SOA serial?
<reallymemorable>
gchristensen: does that mean my card's name is `lo`?
<gchristensen>
no, that is localhost
<gchristensen>
you don't by chance have an android phone, do you? I suspect the installer doesn't have drivers for your wifi card by default
sondr3 has quit [Quit: WeeChat 2.2]
<reallymemorable>
no i have an iphone
<reallymemorable>
i am installing this on a 2013 macbook air
<gchristensen>
ah. hrm. it is likely you'll need a custom-built install ISO. this isn't complicated, but I can't do it for you just now ... someone else might be able to -- you probably need th broadcom kernel module
fusion809 has quit [Quit: Leaving]
<reallymemorable>
ah
<reallymemorable>
What kind of laptops do people tpyically run NixOS on?
<reallymemorable>
Also would a 2018 MacBook Pro require the same custom kernel?
<gchristensen>
I've run it on a few macbook pros (2013 was a good year) and dell xps
<gchristensen>
yeah. it really isn't a big deal to make the custom iso, just can't do it tonight :/
eadwu has quit [Quit: WeeChat 2.3]
<reallymemorable>
What would be the best way to follow up?
<reallymemorable>
Find you on here another day?
<gchristensen>
yeah, that would work -- 30hrs from now for example. maybe other people could do it, too
<reallymemorable>
ok much appreciated
<disasm>
2018 has T2 chip. I don't even think you can boot linux on it anymore because of the TPM.
<reallymemorable>
oh man thats unfortunate
<reallymemorable>
i was hoping to have a dualboot MacBook Pro
<disasm>
2013 macbook air works well
<reallymemorable>
ok well thankfully i also have one of those
<disasm>
I have a 2011 macbook air and 2014 macbook pro both running nixos
<reallymemorable>
the macbook air is kind of beat up
<gchristensen>
maybe disasm can make such an ISO :)
<sb0>
clever: the jobset declaration itself is also hit by the bug, which wiping the DB didn't fix
<DrLambda>
When trying to upgrade nixos I get "HTTP error 200 (curl error: Stream error in the HTTP/2 framing layer);" is there maybe a way to use http/1 instead?
<clever>
sb0: and in the project settings, you give it the git url, and the relative path, of that spec.json
<sb0>
clever: yep, I did that
<clever>
sb0: spec.json tells it how to build hydra-project.nix
drakonis1 has joined #nixos
<clever>
sb0: hydra-project.nix then generates a set of jobset->spec.json, to decribe how to build everything else
<sb0>
clever: wait, the project settings need to point to a .json file and NOT a .nix file that creates a spec.json? the documentation suggests otherwise
<disasm>
reallymemorable: image building
pbb has joined #nixos
<clever>
sb0: correct, the settings must point to a root spec.json file, like the one i linked above
eadwu has joined #nixos
<reallymemorable>
thank you so much
<sb0>
clever: though, hydra should have reported a JSON parse error, and not make the evaluation get stuck
<disasm>
reallymemorable: it has zfs stable in it as well, if you want unstable let me know and I'll abort and start over
<clever>
sb0: that json error is likely somewhere in the journal logs
<reallymemorable>
I am new to this world
<sb0>
clever: changed to a .json at the root, same hydra behavoir
<reallymemorable>
so im guessing i want stable
<clever>
sb0: hydra doesnt update it immediately
<sb0>
clever: I updated the project settings + did a "force evaluation" on the .jobsets - isn't that supposed to make hydra update it?
<clever>
sb0: then it could also be cargo or rust, one sec
<clever>
--option substituters ""
<clever>
sb0: add that to the `nix-store -r --dry-run /nix/store/0gaksckwpi6cwkmkncd8pm5mhbr7z837-conda-artiq-board-kasli-tester.drv --option substituters ""`
<clever>
then youll see 2 new .drv files, for cargo and rust
<clever>
show-derivation each of them
<reallymemorable>
disasm: when do you think the ISO will be ready?
<clever>
sb0: you dont have the binary cache on, so hydra has to build a copy of rustc
<clever>
sb0: but rustc requires the "big-parallel" feature on a build slave
<clever>
and the default localhost slave lacks that feature
<clever>
sb0: services.hydra.useSubstitutes will allow hydra to just download rustc from cache.nixos.org, and un-hang it
daniele- has joined #nixos
<sb0>
clever: it worked!
<sb0>
clever: thank you so much!!
<daniele->
hi!
<clever>
sb0: its possible that you already had rustc, from pre-hydra testing, and when nixpkgs changed, it had to get a new rustc
<daniele->
is the setup of the default firewall somehow conditional? depending in how i bring up the nixos container it installs or not the firewalling rules
<sb0>
yep, I did a lot of manual rust compilations on that machine
<daniele->
in both cases the container is instantiated with systemd-nspawn
<daniele->
in one case manually from the command line and in the other through machined
<clever>
that removes the roots, so a normal `nix-collect-garbage` (with no flags) can then delete it
sigmundv__ has quit [Ping timeout: 246 seconds]
<wedens>
hm. ok, I can probably write some script that uses nix-env. thanks
<wedens>
does `nixos-rebuild test` add new generation?
<wedens>
what I'm trying to achieve is to build and activate new configuration without spawning tons of generations when I'm doing lots of small tweaks
<clever>
wedens: i try to test such stuff without nixos-rebuild
<clever>
test also made no new generation
<clever>
wedens: are you testing a normal package, or a nixos module?
<wedens>
a module
<clever>
wedens: then youll want to use a mix of `nixos-rebuild build` and `nixos-rebuild test`
iyzsong has quit [Ping timeout: 258 seconds]
<wedens>
clever: `nixos-rebuild test` builds a configuration, doesn't it? or do you mean something else?
<clever>
wedens: test will build and activate, but not make a generation
<clever>
and /boot is NOT updated, so all changes are lost at reboot
<wedens>
I should probably stick with my initial plan of "amending" previous generation, I think. with `nixos-rebuild test` it'll be too easy to forget to "commit" :)
reallymemorable has quit [Ping timeout: 268 seconds]
<{^_^}>
#19741 (by cko, 2 years ago, open): Maven builds are not documented in nixpgs manual
<colemickens>
apparently there's a new kid on the block
daniele- has quit [Client Quit]
slack1256 has joined #nixos
lejonet has joined #nixos
<wedens>
I don't think you gain much by using nix for managing java dependencies. native requirements are not that often in java-land. perhaps that's a reason why java infrastructure in nix is not being actively worked on
daniele- has joined #nixos
<colemickens>
I guess it just seems like the fastest route for me to get it packaged
<colemickens>
and I want it packaged to be able to write a nixos module and build a VHD for my dad for an OpenHAB VM that he doesn't have to really touch
<wedens>
colemickens: perhaps the simplest way is to use openhab docker image
daniele- has quit [Quit: daniele-]
FRidh has joined #nixos
daniele- has joined #nixos
drakonis has joined #nixos
<{^_^}>
[nixpkgs] @vcunat pushed 12 commits to release-18.09: https://git.io/fh5a7
Drakonis__ has quit [Ping timeout: 258 seconds]
ninjin has quit [Ping timeout: 256 seconds]
slack1256 has quit [Remote host closed the connection]
_kwstas has joined #nixos
_kwstas has quit [Client Quit]
drakonis1 has quit [Quit: WeeChat 2.3]
<{^_^}>
[nixpkgs] @vcunat pushed commit from @delroth to staging-next « unzip: patch CVE-2018-18384 »: https://git.io/fh5ah
reallymemorable has quit [Ping timeout: 246 seconds]
random_yanek has joined #nixos
random_yanek has quit [Max SendQ exceeded]
random_yanek has joined #nixos
Ralith__ has quit [Remote host closed the connection]
<srhb>
hyper_ch: Just like on any other OS.
<hyper_ch>
srhb: I've never used docker
<srhb>
hyper_ch: Time to read a manual then :)
dermetfan has joined #nixos
<cyris212>
Does someone have hunspell working with an english dictionary?
daniele- has quit [Quit: daniele-]
<tilpner>
hyper_ch: The installation process is different, but the wiki explains it, and you would easily be able to guess it from "man configuration.nix"
<dramforever>
TIL 'man configuration.nix'
<dramforever>
tilpner++
<{^_^}>
tilpner's karma got increased to 17
<cyris212>
Turns out you need to set DICPATH to your nix-profile for it to work.
<cyris212>
tilpner: that's awesome \o/
winem_ has quit [Ping timeout: 250 seconds]
<tilpner>
Maybe it should be advertised on the options website
das_j has quit [Remote host closed the connection]
das_j has joined #nixos
_kwstas has joined #nixos
babic has joined #nixos
_kwstas has quit [Client Quit]
_kwstas has joined #nixos
_kwstas has quit [Client Quit]
init_6 has quit []
sour has joined #nixos
<wedens>
cyris212: there is `dicts` in hunspell derivation
<wedens>
which provides DICPATH for hunspell binary
<wedens>
and `hunspellWithDicts` is hunspell with (all?) dictionaries
<sour>
hi all
drakonis_ has joined #nixos
jomik has joined #nixos
<dramforever>
hi sour
ottidmes has joined #nixos
drakonis has quit [Ping timeout: 252 seconds]
<jomik>
I was curious how you guys that don't use home-manager manage your home configuration? I found that I need to reference a nix package in many places. As an example, my xmonad needs a reference to some scripts in my nix store. Do you just ensure that these are in your path? Do you use an imperative setup for managing your vim/emacs config and shell?
<tilpner>
jomik: I try to generate my configurations with Nix, and that gives me embedding of store paths for free
<dhess>
specifically in package-set.nix. I'm curious how the reference to `self` in the self assignment works
<jomik>
My issue using home-manager is that it can become quite tedious to test configurations - i.e. I want to configure my dunst notifications, but I'd essentially have to use `home-manager switch` with each change. I _could_ just work directly in dustrc.
<sour>
i asked yesterday about this, got some answers, but still didn't fix some problem i got; basically I can't build the X11 haskell package required by xmobar - it fails because it can't find the 'X11 headers'. Looking through the project, I saw that it required X.h, Xlib.h, and some xrandr.h and xinerama.h, among other things
<jomik>
tilpner: How do you generate the configurations then? Do you do it inside /etc/nixos/configuration.nix ?
<jomik>
It gets a bit annoying at times that home-manager and my system config actually are separate. That is, my system config can not actually do `shell = "my fish shell"`, because it is a package derived by home-manager.
<tilpner>
dhess: I'm not sure I understand your question correctly. self is meant to refer to the final package set, and it will only be evaluated when an overlayed item accesses it
<tilpner>
jomik: Yes, I do that from ny NixOS config, with an evil module that allows me to write anywhere (instead of just /etc)
<dhess>
tilpner: Does Nix do dynamic scoping a la Emacs?
<jomik>
Ahhh, essentially what the home-manager module in nixos does.. :O Can you point me to an example?
<wedens>
jomik: in case of dunstrc and other config files that you don't wan't to write in Nix, you can just symlink them and edit as usual
<tilpner>
dhess: I've tried Emacs a few times, and never stuck with it. I'm not qualified to make that comparison
<tilpner>
dhess: Maybe it would help you to play with this snippet reduced to one overlay, and a sample overlay, and see what you can and can't do?
<dhess>
tilpner: ok let me ask a different way. Which `self` is that `self` in the expression bound to?
<jomik>
wedens: Well, I do actually want to write it in nix, it just gets a bit annoying testing changes with home-manager, because it takes a good amount of time.
<tilpner>
dhess: The self on line 6 is bound at line 5, and then used again on line 8
<tilpner>
> let self = { foo = self.bar; bar = 42; }; in self
babic has quit [Remote host closed the connection]
<{^_^}>
{ bar = 42; foo = <CODE>; }
<tilpner>
> let self = { foo = self.bar; bar = 42; }; in self.foo
<{^_^}>
42
<wedens>
jomik: huh? shouldn't take more than ~10 seconds if you don't reconfigure packages
babic has joined #nixos
<wedens>
I'm more annoyed by dozens of useless generations
<tilpner>
jomik: An example of what? My module is available via the NUR
<das_j>
Hey, not sure if this is the correct way to apply for this, but I think nixpkgs commit access would really make my life a lot easier. I maintain some packages now and some trivial changes (like bumping versions) often take a lot of time because they don't get merged. Also, helping to tackle the incredible amount of open PRs could also make your lifes easier. So is there a way to get into the nixpkgs team?
<tilpner>
You can shorten eval time, if you only eval a part of your config, as opposed to system.build.toplevel
<tilpner>
(Down to well below 1s)
<jomik>
wedens: But yeah, generations is annoying too. I am on 268 now
<wedens>
I use home-manager as nixos module, so any change results in switching system configuration. and I can't say it's too slow to be annoying
orivej has quit [Ping timeout: 272 seconds]
orivej has joined #nixos
<jomik>
Ahh, I guess that'd make me more annoyed. I dislike cluttering the boot menu :P
<tilpner>
sour: gist your expression somewhere?
<teto>
jomik: this is a complaint I have too. For programs that accept it, I include a non-declarative file I can change at will. When my changes work out, I move the configuration to home-manager
sour_ has joined #nixos
<tilpner>
Ehh, many generation are good
sour has quit [Ping timeout: 256 seconds]
<dhess>
tilpner: I'm surprised that works.
<wedens>
I'll write a script that "amends" previous generation or I'll just use `nixos-rebuild test` and do ``nixos-rebuild switch` when I'm ready to "commit"
reallymemorable has joined #nixos
<wedens>
so, problem of dozens of meaningless generation is not that difficult to solve
<dhess>
It appears that Nix basically treats an expression that refers to itself like a fixed-point... it does not recurse into the expression
<das_j>
tilpner: Ah thank you, so there are no real new committers until this is cleared up? The discussion doesn't seem to have come to an end, everything is just a broad idea.
<jomik>
Oh, huh, so `nixos-rebuild` test does not make a boot entry wedens ?
<wedens>
yes
<jomik>
teto: Yeah, that's what I figure I'll do as well. I guess I will just use the generated config as a base to then figure out the looks of my dunst notifications, and then when I got the numbers, write it into home-manager...
<jomik>
Tho, I didn't know `nixos-rebuild test` doesn't make a boot entry. I may switch to the home-manager nixos module then :o
<tilpner>
das_j: I don't know, it seems to have worked for kalbasit. Maybe just comment on there? (Or ping ${"inf" + "inisil"})
<teto>
jomik: alternatively you can copy the config and use a program flag to load that alternate config while testing
* inf
big thonk
reallymemorable has quit [Ping timeout: 246 seconds]
<tilpner>
:/
<tilpner>
I tried!
<inf>
haha, don't worry.
<inf>
that's what i already got used to while having quite common 3-letter nickname.
<das_j>
infinisil-m: I'll just ping you here :) Or should I rather add a comment to the issue?
<tilpner>
infinisil: ^
<das_j>
Oh, thanks Weechat for those great completions
<tilpner>
dhess: You'll notice that the source for lib.fix is pretty much the same
<tilpner>
dhess: fix = f: let x = f x; in x;
<tilpner>
All it does is provide a self in a way that's not let-in
<dhess>
it doesn't refer to fix though :)
<dhess>
that's my point
<tilpner>
But it's practically fix. fix is no hugely complicated function
<hyper_ch>
ok, using prebuilt docker images was easier to use than anticipated
<dhess>
yes I understand fix, but the *scoping* is what's odd.
<dhess>
I do not think you could do that in any other "functional" language I'm aware of... except maybe Emacs
<tilpner>
> let a = { x = b.x; y = 42; }; b = { x = a.y; }; in a.x
<{^_^}>
42
<dhess>
Anyway, you've cleared up the issue for me. Thanks
xkapastel has joined #nixos
<sour_>
,locate Xlib.h
<{^_^}>
Found in packages: iv, z88dk, emscripten, xlibs.libX11.dev, perl526Packages.Tk, perl528Packages.Tk, perldevelPackages.Tk
<jomik>
the fix function that tilpner wrote I mean.
<jomik>
Why is the scoping odd btw, dhess ? :O
sour_ has quit [Ping timeout: 256 seconds]
<dhess>
jomik: either I would expect it to complain that `self` within the expression wasn't in scope; or that it would cause infinite recursion since there's no reduction step/base case when evaluating `self` within the expression.
<dhess>
I suppose treating it as a fixed point is a reasonable default since the behavior I expected isn't particularly useful.
<dramforever>
> let self = self; in self
<{^_^}>
infinite recursion encountered, at (string):218:12
<{^_^}>
[nixpkgs] @FRidh pushed 9 commits to python-unstable: https://git.io/fh5rS
<dramforever>
dhess: Maybe you can treat it as a spreadsheet
<dhess>
dramforever: I suspect that has something to do with the `self` in this case not being evaluated eagerly.
<jomik>
dhess: could be. Or maybe you think about it too deeply :P I think this is sort of some of the first that learnyouahaskell tackles. At least indirectly. Through infinite streams. It really sucks evaluating `map` or `filter` on an infinite stream, Haskell can call map/filter and friends on infinite streams, because it is lazy :D
<jomik>
s/stream/list
<sb0>
ah, there's one in the commit log.
<infinisil>
das_j: Commenting on the issue is probably best for now, I'd love to have you on the committer team :)
<das_j>
Thank you, glad to hear that :)
jomik has quit [Quit: WeeChat 2.2]
<dhess>
jomik: There's a subtlety here that I think you're missing (or perhaps it's me and I'm just dense). I understand mapping infinite streams, the `fix` function, etc. What's odd here for me is the fact that a const function is referring to itself. I don't think I've seen a Haskell program that does that.
<dhess>
that Nix expression immediately refers to `self`
<dhess>
what's limiting the evaluation?
<LnL>
the //
<dramforever>
The spreadsheet
<dhess>
Yeah I was trying to think of an equivalent Haskell expression and could not come up with one
<dhess>
I suppose in the Nix case I need to think about how the default.nix is loaded and evaluated
Supersonic has quit [Quit: Bye]
<dhess>
OK I'll just go slink away and think about this some more.
joehh has joined #nixos
Supersonic has joined #nixos
silver has joined #nixos
<dramforever>
I think you need to pay more attention to my spreadsheet analogy :)
<dramforever>
Something like this: put {= A2 + A3} in A1, {= A3 + 2} in A2, {2} in A3
<dhess>
dramforever: I'm considering it. I suppose your point is that the spreadsheet is effectively calling `fix` on the spreadsheet
<dramforever>
Yes
<dhess>
got it. That was helpful, thanks.
<dramforever>
Also you can try to find examples of doing memoization with maps or arrays in Haskell
<dhess>
Yes I think what I was missing here is the context of how the default.nix is being imported by Nix
<dhess>
I was only thinking of it as a self-contained expression
dsiypl4 has joined #nixos
<jluttine>
how can i get the directory of libgfortran.so in nix derivation? i tried ${pkgs.gfortran} but that gives path to some wrapper that doesn't contain lib directory..
<dramforever>
{... (import ./default.nix) ...} -> {let defaultDotNix = (<paste contents of default.nix here>) in ... defaultDotNix ...}
phreedom has quit [Remote host closed the connection]
<dramforever>
(Put a semicolon before 'in')
<jluttine>
that is, it points to /nix/store/....gfortran-wrapper-7.4.0 instead of /nix/store/....gfortran-7.4.0-lib
phreedom has joined #nixos
<dramforever>
jluttine: (gfortran.cc.lib) or (lib.getLib gfortran.cc)
<dramforever>
That's an English 'or', not Nix 'or'
<jluttine>
dramforever: thanks! i tried gfortran.lib and that didn't work. didn't know about this cc
<das_j>
dramforever: So self gives me the fully-modified package set while super gives me the package set with only the modifications of previous overlays?
<dramforever>
Exactly
<das_j>
That clears things up, thank you
<dramforever>
Didn't expect it to help someone this fast. You're welcome.
<das_j>
dramforever: I usually guess which one to use. If there is an infinite recursion, I just use the other one
<aanderse>
das_j: every approve makes it easier for a committer to merge :)
<das_j>
dramforever: Overriding Linux modules was the last thing :x
<das_j>
Not that obvious if you do it for the first time
<sb0>
when a hydra runcommand command prints something, where is it displayed?
<dramforever>
Never seen/done anything like the last thing
<dramforever>
I need to put my thing in the wiki
<das_j>
dramforever: Yes, this way I know where to find it
<_d0t>
hi! Is there any rescue mode in nixos? I have no space left on my drive, and now my system doesn't boot.
janneke is now known as janneke[m]
janneke[m] is now known as janneke_
janneke_ is now known as janneke
<symphorien>
_d0t: iirc add 1 to the kernel cmdline, you will boot in systemd's rescue mode
<ivan>
I'm still curious about this from two months ago, anyone know? <vaibhavsagar> what's the reasoning behind only allowing users in `trusted-users` to use the machine as a remote builder?
<ivan>
afaik non-trusted-users can still build whatever derivations they want from any nixpkgs
<ivan>
(local builds)
<symphorien>
a machine A using B as a remote builder can add arbitrary already built derivations to the store
<symphorien>
of B
<_d0t>
symphorien: doesn't seem to work
<symphorien>
a local normal user must either provide a .drv which results in this output or use a trusted substituter
<dhess>
See the "Turing equivalence?" bit of the README.md
<ivan>
symphorien: I see, thanks
<dramforever>
_d0t: If you can't even boot that, boot into a nixos installation disk, mount all your filesystems at /mnt or somewhere else, and run nixos-enter /mnt
das_j has quit [Remote host closed the connection]
<dramforever>
*shrug* dunno, sorry
das_j has joined #nixos
<_d0t>
Seems like ZFS lost a bunch of files
jtojnar has quit [Quit: jtojnar]
<tilpner>
_d0t: Boot any live system and free up some space
<tilpner>
Which may be more involved with zfs than with non-cow filesystems
<_d0t>
tilpner: here's the thing. It seems that zfs guys changed the encryption property name, and now I can't import the pool using older zfs versions.
<_d0t>
I'm gonna connect that disk to my laptop and try here.
<tilpner>
Okay, now it's a lot more work, but you can build a live ISO with the right zfs version
<hyper_ch>
_d0t: there was a big change in the zfs encryption
<hyper_ch>
_d0t: nixos did provide temporary solution for the legacy encryption
<hyper_ch>
well, the format change happened between in februar 2018.... so I don't think you're affected by that
<hyper_ch>
but you'll still need to free data
<_d0t>
hyper_ch: yeah, i created that zfs pool and datasets in the summer 2018
<hyper_ch>
did you take lots of snapshots and never removed them?
<_d0t>
nope
<_d0t>
but it's a 256G ssd
<hyper_ch>
:(
<hyper_ch>
create live cd with zfs
<_d0t>
I've just connected the disk to my laptop
<_d0t>
zpool import -f worked just fine
<hyper_ch>
run zfs list -S usedbysnapshots -o name,used,usedbydataset,usedbysnapshots,referenced,mountpoint
<hyper_ch>
( I hae that aliased in my .bashrc --> alias zl='zfs list -S usedbysnapshots -o name,used,usedbydataset,usedbysnapshots,referenced,mountpoint' )
<_d0t>
btw, ZoL 0.8.0-rc2 couldn't import the pool but -rc3 workes fine.
<hyper_ch>
pastebin the output of zfs list -S usedbysnapshots -o name,used,usedbydataset,usedbysnapshots,referenced,mountpoint
<hyper_ch>
I wrote myself a little snapshot tool that also emails me when free space falls below x GB
<hyper_ch>
(currently 100GB on my notebook and 200GB on my servers)
<_d0t>
How do I re-import the pool?
<teto>
,locate libuuid.so
<{^_^}>
Found in packages: libuuid, utillinux, libossp_uuid
<srhb>
oddly, latest-finished is pointing to one of those from yesterday, which I don't understand.. (but the channel hasn't been bumped to that either)
<sb0>
I can't get hydra RunCommand to work - the commmand is simply not executed (no output in any log AFAICT, and also I tried having a command that sends something to a netcat server - nothing was received)
<sb0>
srhb: how to make the output more visible though? connect to postgresql and munge with the hydra database?
<sb0>
in the command I mean
<srhb>
sb0: I don't think there's anything helpful in there, you're probably better off writing a custom script that does its own logging of stdout/stderr
<jomik>
Anyone here using Dunst? I get this (A) and (U) text in my notifications, no clue why. Using `dunst -print` it is not there in the message text.
<symphorien>
sb0: you could pipe the script to logger
<delroth>
ekleog: seems like you added meta.timeout = 30; to nixos/tests/opensmtpd back in August. The test has been constantly timing out since January, and I notice it's the only one using this meta.timeout feature. Any thoughts on just removing the timeout on that test?
<delroth>
I think this has mostly shown itself to be a misfeature, given that people haven't noticed and fixed the timeout failures in about a month...
<jomik>
tilpner: unset, so default?
<jomik>
"odd" part is that it is not always the same, it is like (A), (U) and, I think, (AU)..
<tilpner>
jomik: Mine's format = "<b>%s</b>\n%b"
lunabo has joined #nixos
<{^_^}>
[nixpkgs] @FRidh pushed 412 commits to python-unstable: https://git.io/fh5Kn
b has joined #nixos
<lunabo>
Is there a way to easily install software that has not been packaged for Nix using a more traditional "make; make install" or "pip install" workflow?
<lunabo>
My main concern about switching to Nix is that I will need to install some software and I won't have time to figure out how to do it the nice way
<symphorien>
you can compile things normally inside nix-shell
<symphorien>
but it will break on the next gc
<symphorien>
so this is fit for development and quick testing, but not perennial installation
<tilpner>
You can switch to Nix without risks. Switching it NixOS is what takes your compatibility with make install away
<lunabo>
symphorien: tilpner: Hmm thanks, I'll look into those two options more. I once tried to install some Ocaml packages in a nix-shell and think there were hardcoded /bin/mv paths inside the Makefile, causing it to fail
<lunabo>
but it might also be that I didn't set up the nix-shell correctly at the time
rembo10 has joined #nixos
<symphorien>
ah yes hardcoded paths are a problem
_d0t has quit [Quit: Konversation terminated!]
dsiypl4 has joined #nixos
dermetfan has quit [Ping timeout: 245 seconds]
kisonecat has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @marsam opened pull request #55952 → xapian: fix darwin build → https://git.io/fh56S
<wedens>
,tell jomik in dunst (U) means notification has url, (A) means it has actions (which you can access via hotkey). you can disable it via `show_indicators = false`
<{^_^}>
[nixpkgs] @ryantm pushed commit from @davidak to master « dit: 0.4 -> 0.5 »: https://git.io/fh56A
rcshm has joined #nixos
Ariakenom has quit [Read error: Connection reset by peer]
Ariakenom has joined #nixos
<sondr3>
what is that sorcery wedens?
npmccallum has quit [Ping timeout: 250 seconds]
<wedens>
sondr3: huh?
<ptotter[m]>
it seems that services.prometheus.enable installs prometheus 1.8
<sondr3>
the ,tell thingy, is there a way to see all the commands?
<sondr3>
I've also seen ,locate
<ptotter[m]>
any way to get prometheus 2.2.1 instead? nix-env -qa seems to think it should be available
<wedens>
ptotter[m]: services.prometheus.package
fpob has quit [Ping timeout: 255 seconds]
<ptotter[m]>
didn't find that looking at options
<ptotter[m]>
wedens: what's the syntax like?
<ptotter[m]>
I'm on 18.09
<wedens>
,help
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<wedens>
sondr3: ^
reallymemorable has quit [Quit: leaving]
<sondr3>
should've thought of ,help, thanks wedens
<ptotter[m]>
wedens: seems that's not available on 18.09
fpob has joined #nixos
<reallymemorable>
Hi -- last night someone on here with a username like disasm (or similar) was compiling an iso that would work on my macbook air -- I'm not sure if it finished or I just missed his notification.
<reallymemorable>
Is anyone familiar with this person?
<sondr3>
disasm: ping
<reallymemorable>
oh i remembered correctly
<sondr3>
wifi trouble I guess? I installed NixOS on my MBP using a Ethernet-to-USB dongle
<reallymemorable>
yes the wifi card was not recognized
lassulus has quit [Ping timeout: 244 seconds]
<reallymemorable>
i dont have one of those handy and the folks on here last night were saying that it was easy to create an iso that would find the card
midchildan has quit [Ping timeout: 264 seconds]
<infinisil>
Does anybody happen to know how I can make out-of-nixpkgs NixOS options show up in the manual?
<sondr3>
all I had to do was enable the firmware setting
<sondr3>
though I'm not sure if it is the same as on a MBA
heyguyswatup has joined #nixos
simukis has quit [Remote host closed the connection]
<heyguyswatup>
hi guys
<reallymemorable>
srhb: so i am installing on my 2013 air
chross-q has joined #nixos
<reallymemorable>
i separately asked about my 2018 MBP
Ariakenom has quit [Read error: Connection reset by peer]
thc202 has joined #nixos
<heyguyswatup>
I have a question: when I 'nix search sage' there's a three packages; sage, sagemath and sageWithDoc, when I 'nix-env -i sage' it succeeds but the other two are not found? The reason I'm asking is cause the sage package doesn't come with docs or tut
<heyguyswatup>
and I can't install the WithDoc package
<reallymemorable>
sondr3: what year is your MBP?
<heyguyswatup>
and I can't build the doc cause of 'syntax error'
<LnL>
heyguyswatup: it's based on the channel name, so on nixos it's nix-env -iA nixos.hello while for other distributes it's nix-env -iA nixpkgs.hello
<LnL>
-f '<nixpkgs>' is basically what should be the default IMHO
<heyguyswatup>
okay but I'll remember this trick, so to recap; when i nix search the name is the attribute and the thing in the parenthesis is the thing that I write to do a nix-env -i and if multiple packages have the same name in the parenthesis then the first one in the list gets installed but if I want to get a specific one then I can use -f '<nixpkgs>' and -iA
<heyguyswatup>
ahhhhh
<symphorien>
always use -iA
<{^_^}>
[nixpkgs] @aanderse opened pull request #55953 → nixos/beanstalkd: new service for existing package → https://git.io/fh5iW
<heyguyswatup>
hm so I should basically always install via 'nix search' and then nix-env -iA nixos.<name>
<symphorien>
yes
<heyguyswatup>
thanks
<Rusty1>
no nox?
<LnL>
it's much faster because it's not searching through everything for the name
<timokau[m]>
heyguyswatup: `-i` works based on the package name, which happens to be `sage` (for `sage` as well as `sageWithDoc`). `-iA` works with the attribute name. Because of that, you can only specify the difference with `-iA`.
<reallymemorable>
sondr3:thanks!
<timokau[m]>
sagemath only exists because most repos call the sage package sagemath (its an alias)
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fh5i8
Ariakenom has quit [Read error: Connection reset by peer]
Ariakenom has joined #nixos
<heyguyswatup>
okay I understand. I'm super thankful for this channel btw some of these headaches go away so easily when you have a human to talk to <3
lucus16 has joined #nixos
<timokau[m]>
heyguyswatup: Out of interest (I'm the sage maintainer), any particular reason you're installing offline docs as opposed to just using the online version?
<{^_^}>
[nixpkgs] @worldofpeace merged pull request #55525 → tt-rss-plugin-auth-ldap: Use the correct license → https://git.io/fhHpi
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fh5iR
_kwstas has quit [Quit: _kwstas]
<heyguyswatup>
I like having docs for all my programming environments offline in case of airplanes or busses
<heyguyswatup>
Don't like being offline and crippled
<timokau[m]>
heyguyswatup: Alright, makes sense. For what its worth, you can use `function?` for a quick reference even when the docs aren't installed.
<heyguyswatup>
ooh, didn't know about that. Thanks (also for maintaining the package!)
<gregoire>
Hi! I'm trying to configure msmtp on Nixos. Everything works fine but I have to manually copy my config in my home folder (~/.msmtprc) or in the store. I'm lookinng for a way to automatically write msmtprc in the same folder as the msmtp executable in the store. Is there any way to simply do that?
<{^_^}>
[nixpkgs] @rnhmjoj opened pull request #55955 → Fixes for diod → https://git.io/fh5XF
o1lo01ol1o has joined #nixos
<gregoire>
I've tried using writeTextFile but it doesn't seems to work outside the derivation
daniele- has joined #nixos
<Thra11>
I'm trying to fix/update a couple of haskell packages, hsqml and hsqml-demo-morris. I've followed the instructions in nixpkgs/doc/languages-frameworks/haskell-section.md to create default.nix, shell.nix and <package name>.nix in the local sources, and I've added the recommended lines to ~/.config/nixpkgs/config.nix to override hsqml and hsqml-demo-morris. However, it isn't quite working: `nix build
<Thra11>
nixpkgs.haskellPackages.hsqml` builds the local sources, but `nix build nixpkgs.haskellPackages.hsqml-demo-morris` seems to be trying to use the broken hackage sources to build its hsqml dependency instead.
eadwu has quit [Quit: WeeChat 2.3]
<Thra11>
I'm a little confused. The hsqml override clearly works, but it isn't overridden when building hsqml-demo-morris
slack1256 has joined #nixos
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<wedens>
gregoire: you can make derivation that wraps (via `wrapProgram`) msmpt binary with config as a parameter
o1lo01ol1o has joined #nixos
waleee has quit [Ping timeout: 245 seconds]
reallymemorable has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @mroi opened pull request #55956 → glibc: enable cross builds on Darwin → https://git.io/fh513
<wedens>
gregoire: I prefer to use home-manager. you can declaratively configure mail accounts and other stuff or just copy/symlink existing configuration files
eadwu has joined #nixos
phreedom has quit [Remote host closed the connection]
<samueldr>
sorry if this is condescending, but did you *change* stateVersion?
<samueldr>
because you really shouldn't
<samueldr>
,stateVersion lejonet
<{^_^}>
lejonet: Setting stateVersion to the latest release doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you want to update the stateVersion option regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value.
<ryantm>
lejonet: statVersion doesn't affect what version of nixpkgs you are building with
<samueldr>
(yes, the name is... unfortunate :()
<ryantm>
By default nixops uses whatever is on your NIX_PATH
<lejonet>
samueldr: no, I've had that stateVersion for a while
<samueldr>
right, but did you change it to 18.09, while it was another version before?
<lejonet>
and nixpkgs in NIX_PATH is /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs which indeed seems to be stuck at 18.03
<samueldr>
because, if you did, it's not going to do anything, since stateVersion is NOT used to "statefully define the nixos version"
<{^_^}>
[nixpkgs] @dasJ opened pull request #55957 → nixos/icingaweb2: Init the module → https://git.io/fh51u
<lejonet>
samueldr: not impossible, this system has been nuked and restore from backups quite a lot
<ryantm>
Alternatively it's possible that the nixpkgs version in the NIX_PATH was rolled back.
<samueldr>
it's used to know "what's the schema to be expected for the stateful data; to which NixOS 'state version' does it conform?"
growpotkin has joined #nixos
<lejonet>
not impossible, because I'm supposedly on the nixos-unstable channel (which nix-channel --list confirms)
<growpotkin>
Hey hopefully someone can give me a quick refresher; do runtime dependencies get added to "baseInputs"?
<samueldr>
nixos will rebuild using root's "nixos" channel
<infinisil>
growpotkin: Nope
<infinisil>
,runtimeDeps growpotkin
<{^_^}>
growpotkin: In order of preference: Patch source OR ((if it uses PATH -> wrap with new $PATH) AND (if it uses dlopen, (patchelf --set-rpath in postFixup OR wrap with new LD_LIBRARY_PATH)))
<growpotkin>
what would be the appropriate list
<growpotkin>
thank you <3
<growpotkin>
I waas googling a bit and hadn't bumped into it
<growpotkin>
Long live the IRC!
<lejonet>
samueldr: which is indeed pegged at 18.03, I thought I had changed that ages ago
<infinisil>
growpotkin: :)
o1lo01ol1o has quit [Remote host closed the connection]
<lejonet>
samueldr: now stuff is happening, thank you for pointing me in the right direction. I assumed that I had set root to nixos-unstable ages ago, which is why I didn't even bother look at it :)
<lejonet>
samueldr: quite interesting too how little info it gives on a nixops managed machine and the local machine I run nixops from :P
<infinisil>
ma9e: Did you test it?
<samueldr>
it kinda makes sense when you know what it does: it checks components of NIX_PATH and channels, the channel on a nixops deployed machine is irrelevant AFAIUI
<lejonet>
Yeah exactly
<therealwaphire[m>
guys, anybody use taffybar and xmonad here?
<slack1256>
Yep
<lejonet>
even the /etc/nixos directory is irrelevant on a nixops managed machine
<infinisil>
therealwaphire[m: I did at some point, but better just state your problem upfront
<therealwaphire[m>
oh okay
<therealwaphire[m>
so, I'm both a nixOS and Haskell noob
<therealwaphire[m>
installed taffybar through `nix-env -iA nixos.taffybar`
rydnr has joined #nixos
<therealwaphire[m>
Copied the taffybar.example.hs from github and pasted in my .config/taffybar/taffybar.hs
<therealwaphire[m>
but uh, all the import statements keep failing. `Could not find module System.Taffybar`
* infinisil
takes a look
<ddellacosta>
therealwaphire[m: that's the error you get when you start taffybar?
<slack1256>
therealwaphire[m: try 'nix-shell -p haskellPackages.ghcWithPackages (p: [ p.taffybar p.xmonad p.xmonad-contrib])' and try to compile there
<slack1256>
Also read the haskell documentation on nixpkgs
<therealwaphire[m>
<freenode_son "have you tried including it with"> oh, no, I haven't tried this but I've tried `cabal install taffybar` but that failed with multiple errors. as I'm just playing around with it, thought to install taffybar with `nix-env`
<slack1256>
and `man configuration.nix` and search for taffybar
rcshm_ has joined #nixos
<therealwaphire[m>
so here is where I'm a bit confused too
<therealwaphire[m>
I'm not sure if I should install packages per profile / user or invoke them inside a nix-shell
<therealwaphire[m>
I mean, I'm not sure I understand the difference in use-case
<tilpner>
It's subjective, but I prefer users.users.therealwaphire.packages for most packages, and the nix-shell for things I only need once or twice
<rydnr>
Hi! I have a question. I can build a derivation using nix-build, but it fails when I add it to my configuration.nix and do a nixos-rebuild switch. The problem I get is "cc1plus: sorry, unimplemented: 64-bit mode not compiled in". The package I'm trying to build is a library from Firefox called Moz2D. The derivation is just https://github.com/lukego/studio/blob/vm-upgrade/backend/Moz2D/default.nix after converting it to a function.
<ddellacosta>
therealwaphire[m: I have essentially what slack1256 suggested above but in my /etc/nixos/configuration.nix. It works fine. I'm also a relative newbie to NixOS, sounds like I have a month or so on you--my advice to you would be to start there and worry about the right place to put stuff once you have it working
kp__ has quit [Quit: WeeChat 2.3]
<therealwaphire[m>
<freenode_dde "therealwaphire: I have essential"> thanks, man. that's what i'm doing right now. I'm playing around, trying to break things until I'm comfortable enough
<tilpner>
Although even better is not having it in PATH, by e.g. embedding it in a config file exactly where it's needed
<{^_^}>
[nixpkgs] @ryantm merged pull request #53480 → gitAndTools.git-annex: wrap binary on not-Linux to use Nixpkgs' coreutils → https://git.io/fhsk6
<therealwaphire[m>
<freenode_son "have you tried including it with"> also, hey, uh, isn't putting the package in there the same as `sudo nix-env -i` ?
<therealwaphire[m>
cause I've done that, just to be sure
<therealwaphire[m>
same `Could not find module` error
<therealwaphire[m>
<freenode_til "Although even better is not havi"> I'm not sure I understand what you mean "embedding it in a config file where it's needed"
<therealwaphire[m>
I'm not sure I understand what you mean by "embedding it in a config file where it's needed"
<tilpner>
Oh, the Matrix bridge completely breaks pings
<tilpner>
You'll have to do that manually, not by replying to people
<WilliamHamilton[>
question: I want to run a script that requires java, for a game, and I'd like to use steam-run; Since steam-run doesn't include java, I wrote in my config: `(steam.override { extraPkgs = pkgs: [ pkgs.gnome3.gtk pkgs.gnome3.gsettings-desktop-schemas pkgs.hicolor-icon-theme pkgs.jre pkgs.jdk ]; })`
yokyolu has joined #nixos
<WilliamHamilton[>
but I still get the error `java not found`. Could you explain why?
<rydnr>
WilliamHamilton: Have you tried to inject "pkgs.openjdk" instead of "pkgs.jdk"?
<WilliamHamilton[>
rydnr: no, let me try that!
<therealwaphire[m>
<freenode_sla "and `man configuration.nix` and "> and no `taffybar` in there
zupo has joined #nixos
<WilliamHamilton[>
rydnr: no difference, even with openjdk
kp__ has joined #nixos
<WilliamHamilton[>
but what bugs me is: why does this happen. Doesn't `extraPkgs` mean that the libraries are visible in steam fhs?
<WilliamHamilton[>
*chroot?
winem_ has quit [Remote host closed the connection]
<tilpner>
WilliamHamilton[: As you wrote it, this is steam, not steam-run
<tilpner>
Hmm, I've actually not used steam.override myself
<tilpner>
I use steamPackages.steam-chrootenv.override
<tilpner>
Which may be the same, I haven't checked
<tilpner>
But the result of that override is a package that has "steam", not "steam-run"
<rydnr>
WilliamHamilton: I'm far from being an expert, but in the manual it explicitly addresses this (https://nixos.org/nixpkgs/manual/): steam.override { withJava = true; };
<WilliamHamilton[>
rydnr: oh, you're right! Thanks, don't know how I didn't find this earlier!
<rydnr>
Any hint about what is the difference (in terms of gcc config, I guess) between nix-build and nixos-rebuild switch?
<ryantm>
rydnr: I'd expect nix-build to use your user's nixpkgs and nixos-rebuild swtich to use root's
<{^_^}>
[nixpkgs] @FRidh merged pull request #55628 → openblas: fix config breakage introduced for configs using 'false' → https://git.io/fhQwl
<{^_^}>
[nixpkgs] @FRidh pushed commit from @dtzWill to master « openblas: fix config breakage introduced for configs using 'false' »: https://git.io/fh5M2
ddellacosta has quit [Ping timeout: 272 seconds]
<rydnr>
ryantm: but I'm running nixos-rebuild -I nixpkgs=[my-nixpkgs] switch
ddellacosta has joined #nixos
Dagger2 is now known as Dagger
<rydnr>
The only thing that springs to mind is to understand the gcc (and cmake?) wrappers. I've tried with gcc5, gcc6, gcc7, but I always get the "cc1plus: sorry, unimplemented: 64-bit mode not compiled in". I've checked gcc -version both for root and my user is 7.3.0.
<ryantm>
rydnr: Are you having the issue with nix-build or when building nixos?
<rydnr>
when building nixos
<rydnr>
ryantm: nix-build (and nix-shell) builds the package just fine.
<LnL>
how are you building the package vs nixos?
o1lo01ol1o has quit [Remote host closed the connection]
<rydnr>
LnL: ok. It makes sense. The package can be built with the default nixpkgs, but not with the nixpkgs I'm working on.
lassulus_ has quit [Quit: WeeChat 2.2]
<rydnr>
LnL: However, nix-build --arg pkgs '(import [my-nixpkgs] {})' -E 'with import <nixpkgs> { }; callPackage ./default.nix { }' gets built (or at least, it prints the folder in /nix/store)
<LnL>
I think you want with import /path/to/nixpkgs {}; callPackage ./default.nix {}
o1lo01ol1o has quit [Remote host closed the connection]
<rydnr>
LnL: You're right, it's being rebuilt now. Thanks.
b has quit [Ping timeout: 250 seconds]
endformationage has joined #nixos
<ottidmes>
rydnr: what you probably meant to do was: nix-build -I nixpkgs=/path/to/nixpkgs -E 'with import <nixpkgs> { }; callPackage ./default.nix { }' (which is basically prioritizing the paths you supply there over those found in NIX_PATH, i.e. the default value where <nixpkgs> is resolved to)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rydnr>
LnL: It got built correctly, but nixos-rebuild -I nixpkgs=[my-nixpkgs] switch fails (since nothing changed)
<elvishjerricco>
das_j: I don't think you can. Can't build systemd with musl I believe.
Thra11 has quit [Quit: WeeChat 1.9.1]
<das_j>
elvishjerricco: I switched my libc and I'm currently building a system. Let's see what will happen ;)
<{^_^}>
[nixpkgs] @FRidh opened pull request #55962 → buildPythonPackage: initial support for PEP 517 → https://git.io/fh5DB
matijja has joined #nixos
<elvishjerricco>
das_j: I tried it a few days ago and couldn't get systemd building. DBus too, IIRC
<das_j>
elvishjerricco: Heck. But hey, could have been the first freedesktop projects with a sane code base
<{^_^}>
[nixpkgs] @dasJ opened pull request #55963 → icingaweb2Modules: Init all themes I could find → https://git.io/fh5Du
<elvishjerricco>
das_j: The openembedded project seems to have a bunch of patches to get systemd building on musl, but I stumbled upon a systemd issue thread that described a lot of the patches as major security risks
<das_j>
That's sad because I really see a future in musl
<catern>
,locate bin sendmai
hph^ has quit []
<{^_^}>
Found in packages: exim, msmtp, ssmtp, busybox, postfix, nullmailer, system-sendmail
<rydnr>
Is there a way to force the default gcc?
<symphorien>
what are you trying to do ?
Guest79752 has joined #nixos
slack1256 has quit [Ping timeout: 246 seconds]
lassulus has quit [Remote host closed the connection]
<rydnr>
symphorien: I'm not able to build a package using nixos-rebuild switch, but I can build it correctly using nix-build. The error is "cc1plus: sorry, unimplemented: 64-bit mode not compiled in".
lassulus has joined #nixos
<symphorien>
can you use nix-diff to compare the derivation you build with nix-build and that nixos-rebuild fails to build ?
<rydnr>
symphorien: First time I know of nix-diff. I will take a look, thanks!
<ekleog>
<delroth> ekleog: seems like you added meta.timeout = 30; to nixos/tests/opensmtpd back in August. The test has been constantly timing out since January, and I notice it's the only one using this meta.timeout feature. Any thoughts on just removing the timeout on that test? <-- Maybe bump all the timeouts by a factor 2? I had taken (time on my laptop)*4, but looks like it wasn't enough
<ekleog>
the issue is nixos tests have a strong tendancy to just never end when they fail, so we need _some_ timeout to not just wait for the 4hrs limit
<delroth>
ekleog: I don't think hardcoding a number of seconds is a great design though, it's very much dependent on the machine running the tests
<ekleog>
delroth: right, but we need _some_ way of specifying a timeout, and it just happens to be the only one we have right now
<ekleog>
and I'm not sure “short” / “long” / “very long” would be better
<{^_^}>
[nixpkgs] @tadfisher opened pull request #55970 → ktlint: init at 0.30.0 → https://git.io/fh5S2
<ekleog>
ideally machines running the tests would apply a multiplier to the number of seconds
<ekleog>
… actually I was convinced I had added a paragraph to nixpkgs/manual stating this (“please define the number of seconds by keeping it consistent with nixpkgs”), but it looks like I never actually did it
Ariakenom has quit [Read error: Connection reset by peer]
<ekleog>
if you feel like opening a PR fixing either of those issues (timeout too low / manual paragraph missing) feel free to ping me… and feel free to ping me as well if you figure out a better way of handling this :)
<delroth>
ekleog: yeah, but not sure how that multiplier would be defined
Ariakenom has joined #nixos
kvda has joined #nixos
Ariakenom has quit [Client Quit]
<delroth>
ekleog: if the aim here is just to avoid running for 4h with no progress, can we just not use values < 10-15min?
lejonet has quit [Ping timeout: 245 seconds]
<delroth>
30s is just way too prone to overrun for random reasons
o1lo01ol1o has quit [Ping timeout: 255 seconds]
<MichaelRaskin>
Obviously, the multiplier should be calculated based on Static Bash Units… Correcting for load is a harder problem.
fendor_ has joined #nixos
fendor_ has quit [Client Quit]
<ottidmes>
MichaelRaskin: I am looking at making at least pkgs.fetchurl to use secure connections. I know the right curl options to pass, but I am stuck on how to pass the cacert file, sandboxing disallows access to files in /nix/store that are not direct dependencies, but if I were to change fetchurl, won't the derivation produced by it not also change (they would now depend on cacert), any ideas?
<MichaelRaskin>
They are fixed-output, the output path doesn't change on curl updates and won't change on cacert update
<MichaelRaskin>
(too lazy to click and check if the link is to my comment or not)
<growpotkin>
jesus packaging python is murder me :(
<ottidmes>
MichaelRaskin: right, so the .drv would change, but that does not matter, the paths produced will remain fixed due to them being fixed-output
<ottidmes>
MichaelRaskin: then the only thing I need to fix is some infinite recursion, but that should be doable.
o1lo01ol1o has joined #nixos
<ottidmes>
(fetchurl depending on cacert, while cacert's src is defined using fetchurl, should be a matter of only passing it explicitly at a higher level)
<MichaelRaskin>
I think we already have a special early-stage kind of fetchurl anyway
<ottidmes>
correct, fetchurlBoot I believe, its in core pkgs
<ottidmes>
might have to patch that one too, but I will start with pkgs.fetchurl which will capture almost all src in one go (fetchurl, fetchzip, fetchFromGitHub, i.e. the most commonly used ones)
<growpotkin>
I'm trying to install "OCRmyPDF", and needed to add "pikepdf" as a depency, which is failing during the build for homeless-shelter permissions.
Ariakenom has joined #nixos
<growpotkin>
I don't know enough about python to even guess why it is trying to access $HOME, or which files to patch to fix it
<growpotkin>
I need an adult! lol
Ariakenom has quit [Read error: Connection reset by peer]
<ottidmes>
MichaelRaskin: cool, its working :) "successfully set certificate verify locations"
<ottidmes>
MichaelRaskin: still leaves plenty fetchers vulnerable to MITM though, fetchbzr to name one, not sure if I can do anything about that one without reimplementing it (its small, that helps)
<ottidmes>
MichaelRaskin: yeah, I am just going through them one by one, fetchbower seems secure since it imports cacert as dependency and adds it to nativeBuildInputs, fetchcvs seems insecure, but when I check its manpage I see no mention of certificates, so I think its too old software to even support them
<dmj`>
clever: ping-a-ling
<ottidmes>
MichaelRaskin: fetchgit seems to be secure already: GIT_SSL_CAINFO = "${cacert}/etc/ssl/certs/ca-bundle.crt";
joehh has joined #nixos
<MichaelRaskin>
Nice
<ottidmes>
fetchsvn on the other hand has: --trust-server-cert, which I assume is like --insecure of curl, i.e. no checking
<ottidmes>
MichaelRaskin: I guess I can just overlay svn and bazaar and replace them by a wrapper script that filters out those command line arguments
<ottidmes>
since neither have something akin to curlOpts for extra argumetns
<LnL>
ottidmes: hmm, what's wrong with fetchurl? only fetchurlBoot doesn't support https for bootstrapping reasons AFAIK
<ottidmes>
so it supports HTTPS just fine, it just ignores security checks that are not necessary normally due to our use of hashes, but with a prefetch TOFU, you want to trust them only as much as is truly necessary
<scott>
is there an easy way to enter a nix shell for running a 32-bit dynamic exe from itch.io? i can patchelf the dynamic loader (to the x86 version of the loader), but it's still looking for a small set of 32-bit versions of dynamic libs (dl, pthread, rt, stdc++, m, gcc_s, c), i believe
<scott>
(i'm on 64-bit nixos-unstable)
<LnL>
ottidmes: well with support I was expecting SSL_CERT_FILE, not --insecure :)
<ottidmes>
LnL: its actually using CURL_CA_BUNDLE, but due to --insecure it gets ignored (not that it is set in sandboxing normally anyways)
<ekleog>
delroth: iirc the test passed in <5s on my laptop, so I assumed that 30s was already way too long. Sounds like I was wrong, and as it's still in kernel-land at the timeout a *10 or *20 multiplier for the current value would likely make sense…
<ekleog>
Now, tbh I'm seriously thinking that the build machines are being overloaded with jobs, and this makes them slower than they should be due to too many cache invalidations
<growpotkin>
rydnr: do you mean a specific version?
<ekleog>
maybe their job number should just be lowered… anyway, that's off-topic, and a *20 on all meta.timeout values would likely make sense
<emily>
what's the correct/easiest way to disable the annoying behaviour where system boot/gettys/etc. wait for dhcp?
<reygoch>
Is there some PPA that I can use to install Nix package manager on Ubuntu? I haven't found anything like that.
Thra11 has quit [Ping timeout: 246 seconds]
o1lo01ol1o has joined #nixos
<LnL>
ottidmes: actually fetchurlBoot uses the builtin so https works fine, no http bootstrapping trickery necessary
<LnL>
reygoch: it's preferred to manage a nix installation with nix itself instead of another package manager
<LnL>
there are some things that won't work for an apt/rpm install
<reygoch>
LnL: I'm new to nix. Once I nix manually, I suspect there's some command for updating nix?
<reygoch>
LnL: Is that correct?
<LnL>
exactly, nix can upgrade itself :)
<reygoch>
cool
<callahad>
Hm. My terminal emulator won't start under wayland because it can't find libEGL.so. Not quite sure where to start for fixing that. Should be reproducible with: KITTY_ENABLE_WAYLAND=1 nix run 'nixpkgs.kitty' -c kitty
<reygoch>
LnL: What about the cleanup? If I want to remove it completely? Is there a command for that too?
<callahad>
Suggestions for first steps figuring this out?
<LnL>
reygoch: for the default install it's rm -rf /nix
<growpotkin>
dude I have so many issues with my wifi as well
<growpotkin>
if I close the lid on my laptop I have to restart :( I've spent so many hours trying to debug it lol
<Nivpgir>
I see no wifi with if config, and I'm getting some errors in dmesg involving the wifi firmware
<ddellacosta>
so I'm trying to figure out how to use my own Haskell projects in NixOS, and I'm following these instructions: https://nixos.org/nixpkgs/manual/#how-to-build-a-stand-alone-project ...however I set ghc to be ghc844 instead of what is in that example. But I was surprised that it sees to be compiling GHC from scratch instead of using a binary package. What did I do wrong?
<growpotkin>
have you checkout out the nix-hardware repo?
<growpotkin>
ddellacosta: you aren't doing anything wrong
<growpotkin>
ddellacosta: it is intended to do that to avoid versioning conflicts; this allows you to have separate non-conflicting collections of each GHC version
<ddellacosta>
growpotkin: I see, thanks for the explanation
<growpotkin>
ddellacosta: it does eat a lot of disk space though
<growpotkin>
ddellacosta: you can periodically use "nix-collect-garbage" to clean things up
<ottidmes>
growpotkin: I had issues with my ethernet on my desktop, after coming out from suspend, with any kernels > 4.9, I had it fixed by configuring the driver with an option, but the generic fix is to just reload the kernel module after coming out of sleep
<ddellacosta>
growpotkin: yeah mostly I'm cranky because it's taking so much time. I hope it won't do this every time I set up a new project with the same ghc
<ddellacosta>
but I guess it makes sense if it has different dependencies per project
o1lo01ol1o has joined #nixos
<callahad>
yorick: Thanks, that appears to work for me!
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<growpotkin>
ottidmes: I think I had tried some power management scripts to no avail
<growpotkin>
ddellacosta: yeah it is time consuming. Whenever I clear out my garbage I usually rebuild GHC844, GHC863, and GHCHead overnight with my favorite modules so that I don't have to wait when I'm working.
<ddellacosta>
growpotkin: gotcha. Thanks for the tip, I'll keep that in mind
<LnL>
ottidmes: wonder that was added then, I was looking at the prim_fetchurl implementation apparently
<growpotkin>
ddellacosta: you can add them to your config if you want, but I find that when someone breaks a package it will halt my whole system build which I don't want
<ddellacosta>
ah, right
<Nivpgir>
ottidmes: OK, this seems relevant, my problem is I can't do a nixos-rebuild without internet, is there a workaround for this? For the record, I can reboot to a working arch linux which I used to install nixos in the first place, if that helps...
rycwo has quit [Ping timeout: 250 seconds]
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<growpotkin>
ddellacosta: I keep 863 in my config though which is stable (just don't do head or some of the older ones)
<growpotkin>
Nivpgir: You need to move files over from another machine, or fall back to a state where your internet worked
<ottidmes>
Nivpgir: easiest is to just reinstall
<growpotkin>
If you can still boot arch you can build from there
<ddellacosta>
growpotkin: so, just to double-check my understanding--once you have your set of GHCs built with the standard set of modules you use, next time you create a new project using cabal2nix, it won't rebuild GHC (assuming you don't introduce any new dependencies)?
<growpotkin>
ddellacosta: correct the packages should persist, so you will only need to build things that you are missing.
<ddellacosta>
growpotkin: thank you! This is very helpful for my understanding
<growpotkin>
ddellacosta: The only caveat I would add is that cabal2nix takes versioning restrictions very seriously; so whenever possible let them overlap with your other work.
<ddellacosta>
okay, I'll keep that in mind
<growpotkin>
ddellacosta: do you have a shell file that you use for little projects yet? Cabal2nix is kind of cumbersome for little things
xkapastel has joined #nixos
o1lo01ol1o has quit [Ping timeout: 244 seconds]
<Nivpgir>
Hmmm, I'll try to get ethernet somehow and if that fails I'll reinstall
<Nivpgir>
BTW if I just run nixos install from arch again, will it work? Or will it ruin something?
<ddellacosta>
growpotkin: no, I'm still just trying to figure out a sane flow on NixOS--I'm only really getting into it now. Up until now I've been using `cabal new-*` for Haskell work (before NixOS I mean)
<ottidmes>
growpotkin: you can use callCabal2nix to do cabal2nix on the fly within your shell.nix, I use that and only call cabal2nix when actually "releasing"
<growpotkin>
Nivpgir: it will work if you tell it the proper root directory
<growpotkin>
Nivpgir: that is exactly how I would try to do it.
<ottidmes>
LnL: yep, why its being done was in the comment of the file I linked, its perfectly reasonable, and probably even a feature, but it would have been better if it was behind an option, now I have to do workarounds to undo the disabling of TLS verification
<growpotkin>
I just use that little shell for most things haha. I only really use cabal2nix when I'm packing something
<ddellacosta>
growpotkin: thanks!
<growpotkin>
no sweat, you can change the version at the top (or as a nix-shell arg). Add haskell modules in the top list. Add other executables/lib dependencies in the lower bit.
<growpotkin>
dellacosta: because the haskell packs are in a separate list it's easy to pass them to haddock or something.
<ddellacosta>
gotcha
<growpotkin>
(this is just my way, It's served me well for the last year; but cabal2nix might not be as much as a hassle as I think) haha
o1lo01ol1o has joined #nixos
<ddellacosta>
growpotkin: totally understand. In any case seeing how someone else does it is super helpful for my own understanding, even if I end up tweaking things or trying other approaches later
<cnmne>
hello, I'm getting an end-of-file error after running `nix-env -i hello` with a fresh multi-user install on macos 10.14.3; output with verbose=5: https://paste.debian.net/1068496
<LnL>
ottidmes: I think I know how to handle the builtin:fetchurl part, if you're not familiar with nix/c++
blumenkranz has joined #nixos
<ottidmes>
LnL: the only thing I could think of is doing the verification beforehand, so by all means, if you have some way that does not require any rebuilds, I am all ears :)
Nivpgir has quit [Ping timeout: 256 seconds]
<infinisil>
cnmne: Does `journalctl -u nix-daemon -f` output anything when that happens?
<LnL>
these are all fixed outputs so we can change the derivations as long as the name is the same
<ottidmes>
LnL: I just checked all the other fetchers and am now implementing workarounds for them, that still leaves the builtin fetchers and 4 fetchers I could not figure out whether they do TLS verification (they do not explicitly disable it either)
copumpkin has joined #nixos
<cnmne>
infinisil: I don't have journalctl, but i can try running the equivalent commands in launchctl
o1lo01ol1o has quit [Ping timeout: 258 seconds]
<ottidmes>
LnL: how would that work? the verification has to be done while it produces those fixed-output paths
<blumenkranz>
Hello. I would like to fork this NixOS module: https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/programs/firejail.nix in order to make some tests on my computer before submitting a patch. Do you happen to know if the process is the same as overriding packages, or do I have to read some more specific guide?
<ottidmes>
blumenkranz: you can override modules by disabling the default one, for example: disabledModules = [ "programs/command-not-found/command-not-found.nix" ]; that way you can sort of override an existing module, by supplying a module with the same options but your own config implementation of them
<{^_^}>
nix#2523 (by periklis, 14 weeks ago, open): darwin: nix-daemon crashes due to OBJC_DISABLE_INITIALIZE_FORK_SAFETY
kvda has joined #nixos
stigo has joined #nixos
<cbarrett>
Hello. I'm trying to use Digital Ocean's newish support for custom images to load NixOS <https://www.digitalocean.com/docs/images/custom-images/overview/> They dont support ISOs however. I was wondering if any other image formats are generated by hydra?
blumenkranz has quit [Quit: WeeChat 2.3]
<cnmne>
LnL, infinisil: reading the system.log with console shows a bunch of forks calling and crashing with objc
eadwu has quit [Quit: WeeChat 2.3]
<cnmne>
so that does seem like the issue
<daniele->
fuxk xfinity! i just discovered that they throttle my connection in upload
<daniele->
all connections that tranfer more than 1Mb in upload are throttled to 256 kb/s
reygoch has quit [Quit: Leaving]
<infinisil>
daniele-: so 999kb/s works? And it also works if you use multiple connections?
<daniele->
infinisil: the first Mb goes at ~2Mb/s the remaining is throttled at ~256kb/s
<infinisil>
Ah right, i misread as 1Mb/s
<daniele->
i haven’t tested extensively but i believe it is per tcp connection. i don’t know how it would go for parallel connections
o1lo01ol1o has joined #nixos
<infinisil>
That sure is not nice of them. (Also it's offtopic for #nixos :P, there's #nixos-chat for such things)
schjetne has quit [Ping timeout: 245 seconds]
Guest75765 has joined #nixos
__monty__ has quit [Quit: leaving]
<Guest75765>
OK well, I did that, and something went wrong with the file system, files started getting corrupted, so I formatted everything and now I'm installing again, I'll report back tomorrow, thanks!
Guest75765 has quit [Client Quit]
Nivpgir has joined #nixos
o1lo01ol1o has quit [Ping timeout: 255 seconds]
<Nivpgir>
Whoops I was logged in as guest
<Nivpgir>
OK well, I did that, and something went wrong with the file system, files started getting corrupted, so I formatted everything and now I'm installing again, I'll report back tomorrow, thanks!
<clever>
Nivpgir: nix can detect corrupt files without a reinstall