<ashkitten>
#102595 was merged without making sure it builds. it does not build because the patches that say "can be removed on next release" were not removed
dstzd has quit [Read error: Connection reset by peer]
dstzd has joined #nixos-dev
worldofpeace has joined #nixos-dev
ryantm has joined #nixos-dev
philipp[m] has joined #nixos-dev
symphorien[m] has joined #nixos-dev
Ox4A6F has joined #nixos-dev
emily has joined #nixos-dev
rycee has joined #nixos-dev
maralorn has joined #nixos-dev
regnat has joined #nixos-dev
aterius has joined #nixos-dev
mkg20001 has joined #nixos-dev
roberth has joined #nixos-dev
nh2[m] has joined #nixos-dev
siraben has joined #nixos-dev
colemickens has joined #nixos-dev
bbigras has joined #nixos-dev
bk1603[m] has joined #nixos-dev
matthewkenigsber has joined #nixos-dev
alexarice[m] has joined #nixos-dev
thefloweringash has joined #nixos-dev
jonge[m] has joined #nixos-dev
DamienCassou has joined #nixos-dev
aanderse has joined #nixos-dev
michaelpj has joined #nixos-dev
Dandellion has joined #nixos-dev
Ericson2314 has joined #nixos-dev
timokau[m] has joined #nixos-dev
jtojnar has joined #nixos-dev
JJJollyjim has joined #nixos-dev
Irenes[m] has joined #nixos-dev
danielrf[m] has joined #nixos-dev
puzzlewolf has joined #nixos-dev
Valodim[m] has joined #nixos-dev
domenkozar[m] has joined #nixos-dev
ma27[m] has joined #nixos-dev
rnhmjoj has joined #nixos-dev
zowoq[m] has joined #nixos-dev
treed[m] has joined #nixos-dev
dstzd has quit [Client Quit]
dstzd has joined #nixos-dev
<infinisil>
das_j: nope
<mkaito>
in the test API, there's a method to copy from the vm, a method to copy TO the vm, but no method to copy between VMs. any way to do that, other than ssh?
<das_j>
nice
FRidh has quit [Ping timeout: 240 seconds]
FRidh has joined #nixos-dev
<siraben>
Can I make `/nix` be a symlink that points to a different volume?
<siraben>
On my reMarkable tablet there's not enough space on the root partition to install Nix
<qyliss>
I believe Nix will warn about that by default
<siraben>
And without it having to rebuild everything?
<qyliss>
yes
<qyliss>
although the reason it warns is that if anything resolves symlinks on a path in its build directory, it'll get the non-/nix path
<siraben>
How do I do that? I have the installation script `install` ready
<qyliss>
the install script is probably not going to like this
FRidh has quit [Ping timeout: 268 seconds]
<qyliss>
if I were in this situation I'd build Nix from source on the other volume, set up /nix as a symlink pointing to an empty directory, install Nix with the built-from-source Nix, then remove the built-from-source Nix
FRidh has joined #nixos-dev
<siraben>
build Nix from source on the device?
<qyliss>
yeah
<siraben>
it's an armv7l, gonna be awfully slow
<qyliss>
at a custom prefix on the other volume
<qyliss>
well you could built it on something else as long as you got the prefix right
<qyliss>
that's not important
<qyliss>
you could _try_ running the installer with /nix being a symlink to an empty directory, but I wouldn't expect it to work
alp has quit [Ping timeout: 272 seconds]
<siraben>
Hmm
<siraben>
A compromise might be to cross-compile statically as much as possible for now.
<siraben>
How can the performance of Nix be improved? What's the current bottlenecks?
alp has joined #nixos-dev
<regnat>
siraben: You should be able to install Nix with /nix already existing. So if you make it a symlink (or better a bind-mount) before running the installer it should work
<siraben>
regnat: Should I hardlink it?
tilpner has quit [Quit: tilpner]
<qyliss>
you can't hardlink a directory
<qyliss>
you also can't hardlink between volumes
<siraben>
`error: the path '/nix' is a symlink; this is not allowed for the Nix store and its parent directories`
<siraben>
`./install: unable to register valid paths`
<regnat>
siraben: nah you can't hardlink it indeed, but you can mount a directory (like you'd mount a partition) with mount --bind src dest
<siraben>
heh i only have 500 MB of ram, need to make a swapfile
tilpner has joined #nixos-dev
<domenkozar[m]>
ugh
<domenkozar[m]>
error: /nix/store/y41vgrya42wlzq8zpywkccaiknzpcraz-xz-5.2.5/lib/liblzma.so: error: undefined reference to 'pthread_sigmask', version 'GLIBC_2.32'
<domenkozar[m]>
there we are again, glibc from 20.09 and master don't work together
<siraben>
Swapfile added! Hopefully another GB lets me upgrade nix
ehmry_ has joined #nixos-dev
ehmry_ is now known as ehmry
<siraben>
regnat: Will the mount survive a rebbot?
<siraben>
reboot*
<regnat>
siraben: nope, you need to run each at each boot
<regnat>
s/each/it
<siraben>
I just aborted the upgrade to nix 2.3.8 since it would have spent a lot of time compiling things, time to learn about remote builders I guess?
<sphalerite>
siraben: I used a bind-mount to mount nix from the data partition to /nix
<sphalerite>
siraben: and cross-compiled everything I use on the remarkable basically
<sphalerite>
I don't remember how I made the bind-mount persistent though, maybe by editing /etc/fstab
<siraben>
sphalerite: oh, another remarkable user? nice!
<sphalerite>
:)
<sphalerite>
I think I mostly used nix to get rsync on it, so I can back my notes up
<sphalerite>
I love how hackable the software is, not so much how hackable the hardware is…
abbe__ has quit [Quit: “Everytime that we are together, it's always estatically palpitating!”]
FRidh has quit [Ping timeout: 260 seconds]
FRidh has joined #nixos-dev
<red[evilred]>
Thanks - I guess I just need to keep a running count so I know if any have been merged / rejected then I guess
<red[evilred]>
not a huge deal
<red[evilred]>
since nobody else is going to open a PR in my name - shouldn't be hard to track.
alp has quit [Ping timeout: 272 seconds]
abbe_ has joined #nixos-dev
<siraben>
red[evilred]: yeah contributing to nixpkgs was when i started having to juggle dozens of PRs simultaneously, I turned on GitHub email notifications as result and it's quite good
<red[evilred]>
I did that but turned on all notifications because I wanted to have automation to track all the things
<red[evilred]>
it looks like I may need to set up a separate github account with a separate EMail address to separate the two I guess
<red[evilred]>
I mean I could write some code to follow the event stream for all my issues / prs
<red[evilred]>
and it's easy to do
<red[evilred]>
I just didn't want to reïnvent the wheel if someone had done it alreadu
thibm has quit [Quit: WeeChat 2.6]
<red[evilred]>
Question - Do Release Managers continue to manage their release until EOL?
<red[evilred]>
or are they just responsible for it until release day?
<siraben>
When trying nix-copy-closure I get `bash: nix-store: command not found` as a response, however when I ssh into the device, running nix-store works
<gchristensen>
try `ssh foo nix-store`
<cole-h>
Sounds like the remote machine isn't sourcing the profile to get Nix in PATH on non-interative sessions
rajivr has quit [Quit: Connection closed for inactivity]
<siraben>
gchristensen: yep, `bash: nix-store: command not found`
<gchristensen>
cole-h's got the right idea :)
<LnL>
I really wonder if nix should just try to fallback to ?remote-program=/nix/var/nix/profile/default/bin/nix-store by default
orivej has joined #nixos-dev
<LnL>
90% of the time that works
sgrunert has quit [Remote host closed the connection]
<red[evilred]>
the only thing that scares me about fallbacks is taht you lose a lot of predictability that way.
<red[evilred]>
I guess it depends on what is more important to (the wider) you.
<red[evilred]>
That it work "Every Time"™, or that every execution of the same input results in the same output.
alp has joined #nixos-dev
alp has quit [Ping timeout: 272 seconds]
justan0theruser has quit [Ping timeout: 260 seconds]
<siraben>
cole-h: Hm, which profile is it sourcing to get Nix in PATH?
<andi->
For starters we could document the remote-program flag somewhere other than just the source code.
justan0theruser has joined #nixos-dev
alp has joined #nixos-dev
<siraben>
` NIX_SSHOPTS="source .profile;" nix-copy-closure` worked
<sphalerite>
oh goodness
<gchristensen>
beautiful
<siraben>
ok, now to corss-compile Nix and upgrade it on the tablet. Maybe I can get remote builds to work so `nix-build` on the tablet would automatically forward it to my computer.
<siraben>
If I'm understanding that correctly
<rycee>
Howdy, anybody with Nix 3 insights around? I'm curious if the `nix profile {info,install,remove}` commands are considered stable or experimental? In HM we use `nix-env -i`, which apparently is causing errors like "'/nix/var/nix/profiles/per-user/user/profile' is incompatible with 'nix-env'; please use `nix profile` instead" for users who use Nix 3…
<rycee>
So I'm wondering if it would be considered safe to use the `nix profile` commands inside an if statement that checks for `nix --version >= 2.4`.
<cole-h>
All of nix(1) is experimental, FWIW.
orivej has quit [Ping timeout: 272 seconds]
<rycee>
Yeah, that is my impression as well, considering `nix --help` says "Note: this program is EXPERIMENTAL and subject to change." :-) But I was hoping that the profile sub-command might be more stable?
<rycee>
Considering that it seems incompatible with the corresponding stable nix-env command.
<gchristensen>
cole-h: to confirm: you're hacking on thaht?
<cole-h>
Trying, but it doesn't seem like SpotmarketRequestInstanceParameters has an IPXEScriptURL parameter. Adding that would require patching packngo
<gchristensen>
*deep sigh*
<gchristensen>
can you PR that to packngo, and propose a (obviously broken) patch to the terraform provider linking to the PR to packngo?
<{^_^}>
#104484 (by danderson, 2 days ago, merged): influxdb2: init at v2.0.2
<danderson>
thanks chrome for updating the rendered page but not the URL. Real helpful.
<cole-h>
Not just Chrome. Happens to me on Firefox as well.
<samueldr>
>> cannot read '/nix/store/aip17cbq5ywh0xp5hkx4kq98g84zhjzd-source/ui/package.json', since path '/nix/store/zcjs42hsvrfqlp4smhdszranpsai3im6-source.drv' is not valid,
<samueldr>
mkYarnPackage might do something at eval time causing an "IFD"
<samueldr>
,IFD
<{^_^}>
import-from-derivation (IFD) is when you evaluate nix from a derivation result, for example `import (pkgs.writeText "n" "1 + 1")` will evaluate to 2. This is sometimes problematic because it requires evaluating some, building some, and then evaluating the build result. It has been described as "such a nice footgun."
<danderson>
hrm. It shouldn't be, given it's how you're supposed to build yarn stuff in general?
<danderson>
but yay for me discovering a footgun I guess
<samueldr>
I'm looking up the info
<danderson>
oh, hmm. I see. It's reading package.json in order to generate a derivation containing the node deps
<samueldr>
but IIRC one of the yarn2nix has two modes of operations
<samueldr>
yeaj
<samueldr>
yeah*
<danderson>
yeah, there's a mode where I can pre-generate the .nix
<danderson>
so I guess that's what that's for :)
<samueldr>
look at pkgs/applications/networking/instant-messengers/element/element-desktop.nix
<danderson>
okay, this helps, thanks.
<samueldr>
it uses mkYarnPackage, but has the .json and a generated .nix file
<samueldr>
>> # NOTE: this is optional and generated dynamically if omitted
<danderson>
got it
<danderson>
yeah, I saw it was optional and went "well no need to do that then"
<danderson>
so, one more question: why did this fail on ofborg and not locally? Does ofborg toggle a special "be mean about IFD" mode that I could also enable?
<danderson>
retiring committers after 3 minutes of inactivity seems a bit harsh
<danderson>
but hey, you're a true believer or you're not!
<andi->
either you commit to it or not ;-)
<gchristensen>
at least it is a duration, after 1 yard is even weirder. 1 yard of patches?
<danderson>
samueldr: and yeah, buildGoModule requires the sha256 of the "vendored packages" derivation
<danderson>
it would be nice to have some way of making IFD more principled somehow, so that integrating with package managers that have their own locks and hashes doesn't require extra work
<danderson>
otoh, it's a minor extra effort to make nixpkgs completely hermetic, so...