<spacetato>
builder for '/nix/store/wvi4zfq7bcp39jbq5g56g1nj77cd55hb-tmux-3.1c.drv' failed with exit code 1; last 2 log lines:
<spacetato>
ErroSysError~executing '/nix/store/93xwsgby6fchr8a58ckgcm2lf4nkv6fl-bash-4.4-p23/bin/bash': No such file or directory
<spacetato>
anyone seen something like that before?
fendor_ has joined #nixos
jb55 has joined #nixos
fendor__ has quit [Ping timeout: 246 seconds]
jb55 has quit [Remote host closed the connection]
jb55 has joined #nixos
<growpotkin>
Well that's not good
<spacetato>
and I am able to execute that bash without issue, so not sure why it is saying "No such file or directory"
<growpotkin>
The first thing to find out is if "the path to bash does not exist" or if "an unnamed program could not be found at the expected path"
<growpotkin>
I usually expect "/path/to/foo: No such file or directory", but it could also be `bash` complaining that it can't find some random executable or directory
<growpotkin>
can you check that the path provided exists in your store?
<spacetato>
ah, that makes sense
<spacetato>
and yes, it does
<spacetato>
and I am able to execute it from my shell
<growpotkin>
okay so my bet is that bash can't find something, but it's not telling you what it was looking for ( crappy error message )
<growpotkin>
whats the script you are invoking when it complains?
<spacetato>
i'm not sure where that call is happening.. that's the error I see trying to build tmux, without any further context
mbrgm_ has joined #nixos
<growpotkin>
Oh okay, so it's probably in the genericBuilder
<growpotkin>
when you do your build one good place to start is adding `--show-trace` to the nix command to see if it provides any useful info
<spacetato>
nothing additional there; do you know how to get `nix build` to show more than the last two output lines?
mbrgm has quit [Ping timeout: 240 seconds]
mbrgm_ is now known as mbrgm
<spacetato>
also this isn't specific to tmux; i'm unable to build nixFlakes for aarch64 because the `mdbook` build spits out the same "No such file or dir" error
<spacetato>
and nixflakes depends on mdbook
shibboleth has quit [Quit: shibboleth]
<growpotkin>
did you do `nix-build foo --show-trace` already?
andrewrk has left #nixos ["Leaving"]
<growpotkin>
if you aren't getting anything useful there, I would use `nix-shell` to step through the build manually
cosimone has joined #nixos
<growpotkin>
and probably download the `default.nix` and add extra debug prints to the builder if you can't find anything helpful
supersandro2000 has quit [Disconnected by services]
iH8c0ff33 has joined #nixos
supersandro2000 has joined #nixos
cosimone has quit [Client Quit]
<spacetato>
ok, trying to remember which phases I have to run in the shell...
<{^_^}>
#72657 (by cleverca22, 1 year ago, merged): Initial implementation of vc4 cross-compile
<clever>
ali1234: that is a PR where i added a new gcc, to support cross-compiling to vc4/vc6
rajivr has joined #nixos
<clever>
so it acts as an overview of which files are involved in cross-compile setup
<ali1234>
but that's for a "none" system?
<clever>
yeah
tkei has quit [Ping timeout: 245 seconds]
<spacetato>
growpotkin: looks like the error is in the installPhase
<spacetato>
I think..
<spacetato>
everything worked fine up until that point
<spacetato>
and then during installPhase, the first line of output is:
<spacetato>
/nix/store/z5iwwyw6d3gyqc6m8c3ygyahdc8d4ng0-coreutils-8.32/bin/mkdir: cannot create directory ‘/nix/store/kwvqalrv4f29sxkd9ldad39pgzmfs4c8-tmux-3.1c’: Read-only file system
<spacetato>
and there's a bunch of similar errors
<spacetato>
and then the last one is a "No such file or directory"
<spacetato>
:
<spacetato>
/nix/store/z5iwwyw6d3gyqc6m8c3ygyahdc8d4ng0-coreutils-8.32/bin/cp: cannot create regular file '/nix/store/kwvqalrv4f29sxkd9ldad39pgzmfs4c8-tmux-3.1c/share/bash-completion/completions/tmux': No such file or directory
<spacetato>
so I imagine bash is exiting with that last exit code, and that is why I am seeing that message
<{^_^}>
[mobile-nixos] @samueldr opened pull request #247 → asus-dumo: Fix regression from kernel builder changes → https://git.io/JkKfj
ris has quit [Ping timeout: 240 seconds]
lordcirth__ has quit [Remote host closed the connection]
lordcirth has joined #nixos
<pbb>
I keep forgetting this: What is seen as a 'mass-rebuild' and should go to staging? I have a rust update with 3k rebuilds and wondering if it can go on master.
<{^_^}>
[nixpkgs] @lopsided98 opened pull request #104647 → pythonPackages.pykdl: fix build with latest SIP → https://git.io/JkKqU
<zane>
Does the kernelMajor attribute seen in platforms.nix actually do anything? It doesn't seem to be referenced anywhere
<samueldr>
it sure looks like it's not a thing used anymore
spacetato[m] has joined #nixos
<samueldr>
the last use in Nixpkgs was removed in 2018, but might not actually have been useful
<zane>
Hmm, damn. I was hoping it'd make things easy to force a build targeting kernel 2.6
<siraben>
samueldr: how did you determine that? curious
<samueldr>
siraben: `ag` (or ripgrep, or grep) first to confirm it indeed wasn't used
<samueldr>
and then `git log -S kernelMajor` (which takes a hot minute)
<samueldr>
zane: its use really was to distinguish between 2.4 and 2.6
<samueldr>
if you want to know how vintage that was :)
<zane>
oh damn, that was old
<zane>
I need to build a static binary that'll run on 2.6, that's what I was hoping to use it for
<samueldr>
I don't think that it was meant for that
<samueldr>
I think in your case everything will need to be aware about not using too-new features
<samueldr>
you did test and see if `pkgsStatic.whatever` worked before thinking too hard, right?
<zane>
Yep, is SIGKILL'd immediately because of an invalid syscall
<samueldr>
right
<siraben>
samueldr: ah very cool, -S shows the entries that change the number of occurences.
spacetato is now known as spacetato-test
<samueldr>
siraben: I've seen it called the "pick-axe"
<siraben>
Why pick-axe?
<samueldr>
I don't know :)
spacetato[m] is now known as spacetato
<samueldr>
siraben: but if you search git pick-axe online you'll find about -S
<samueldr>
and it's even referenced as such in the man page
<samueldr>
(pickaxe though)
<zane>
Since kernelMajor isn't a thing, what would one use to target an older kernel? Would overriding linuxHeaders do it?
<samueldr>
in theory linuxHeaders are always backwards compatible
<samueldr>
though I don't know if in actuality it is
spacetato-test has quit [Quit: leaving]
<samueldr>
from what I understand, it's likely any of the libs, or the software itself, needs to be aware not to produce *that* instruction that it fails on
<zane>
I just mean to force musl/glibc to only use 2.6 syscalls
<samueldr>
I guess *maybe* features detection from looking at the kernel headers could do that
<siraben>
#! nix-shell -I nixpkgs=https://github.com/NixOS/nixpkgs/archive/82b5f87fcc710a99c47c5ffe441589807a8202af.tar.gz, is it a problem if I can't specify the hash?
<samueldr>
siraben: in practice it shouldn't be a problem, but if someone hijacked github, or the https connection, they could feed anything
<samueldr>
(or if there's a bug in github and it starts serving another commit)
<siraben>
Right. Is there a way to specify the hash?
<samueldr>
I don't think there is using NIX_PATH
<siraben>
Assuming because in general it could be some fork/mirror of nixpkgs which may not be as reliable etc
<siraben>
I see
<{^_^}>
[nixpkgs] @samueldr opened pull request #104648 → platforms.nix: Remove now unused kernelMajor → https://git.io/JkKYE
bbarker has quit [Remote host closed the connection]
<daf93>
It might be a good idea to package nexe for nixos
<__red__>
Great idea! You should!
<{^_^}>
[nixpkgs] @jtojnar merged pull request #103779 → Separate AdobeBlank-Regular.ttf from google-fonts package, moved to google-fonts.adobeBlank → https://git.io/Jk3LF
<{^_^}>
nixos-search#233 (by lheckemann, 18 seconds ago, open): Structured option defaults and examples are rendered as JSON
<Reventlov>
perfect, thank you :)
hlolli__ has joined #nixos
hyper_ch4 has quit [Ping timeout: 272 seconds]
supercoven has joined #nixos
<deadpixels>
Is it possible to setup grub 2 so that it selects the most recently booted OS each time? In the grub docs I read I should set GRUB_DEFAULT and GRUB_SAVEDEFAULT, and in the nix config I see I can assign grub.default, but how do I tell it to save the default? I assume I'll have to pass extra config, but I'm a little hesitant to experiment and break the grub installation :D
<sphalerite>
deadpixels: if you have your config in git and a bootable installer handy, I think experimenting won't hurt :)
<deadpixels>
sphalerite: I guess that's the benefit of using NixOS :)
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
<sphalerite>
deadpixels: Yep! To recover from breaking it, boot into the installer, mount the filesystem, check out the last known working config, and rerun nixos-install.
<typetetris>
Or if it was some mistake in a runnig system, just reboot and select an older entry in the boot menu. (Obviously has drawbacks for servers ...)
<dminuoso>
We intend to switch our haskell project to nix, but are unsure how to tie it into our gitlab ci. So we can just run a gitlab ci runner with nix installed (or even on a nixos), how can we use this as a basis for a binary cache?
sphalerite has quit [Quit: reboot time!]
<dminuoso>
Are there manuals, guides or resources exploring how to build this?
<dminuoso>
I mean naively Id start by having the ci just run `nix build`
<dminuoso>
How does haskell.nix compare to say the nixpkgs infrastructure?
<dminuoso>
I thought we had stuff like callCabal2nix in nixpkgs
<mkaito>
we tried everything, even wrote our own at some point, and now we just use haskell.nix. I'm not personally familiar with the particulars, I try to stay away from haskell myself.
<mkaito>
heresy, I know :P
sphalerite has joined #nixos
<mkaito>
but the rest of my team is also here, they might respond if they have the time
<srhb>
dminuoso: Once the build is done, you get a closure. Personally we serve (the entire store) directly from our builder, which also takes care of signing, but uploading to a cache is basically just a nix-copy-closure on the result path to whereever
<srhb>
dminuoso: Nothing specific to haskell.nix in here.
<mkaito>
yeah, we did that, but ran into constant drive space issues.
<srhb>
Pushing to S3 is fine too, makes no difference.
<mkaito>
if you just want to publish the result, cachix might also work for public caches. I wouldn't directly serve a server's nix store to the public tbh.
<dminuoso>
srhb: Does nix-copy-closure only copy what's needed on the remote side?
<dminuoso>
mkaito: No, this is internal only
<srhb>
dminuoso: At least over ssh. And I assume the same is true for s3.
<mkaito>
yes, it will only copy the runtime graph
<dminuoso>
srhb: I see. So the most naive CI job could just run `nix build ... && nix-copy-closure ...`?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6IJ
<dminuoso>
Well, our deployment requires a little more than just this haskell package.
<mkaito>
wel yeah
<mkaito>
tends to be that way
<dminuoso>
We need to manage a fleet of 8-10 servers, with postgres running on all of them, streaming replication, the haskell server operating on all with configuratoin differences, etc..
<dminuoso>
And firewalling, prometheus exporters, etc...
<mkaito>
sounds like a perfect job for nix
<dminuoso>
So Im not sure what other options than nixops I have here
<simpson>
Pg's the only part of that which isn't easy in Nix.
<mkaito>
yeah, I just stopped trying to use nix for pg tbh
<mkaito>
I just tell nix to put up the server and then manage it manually
<dminuoso>
simpson: what exactly is difficult about it?
<mkaito>
but I don't put my postgres on application servers either
<dminuoso>
mkaito: Our services needs to operate even during network outages, so we quickly arrived at the idea to run a local read-only replica on each shard.
<dminuoso>
(The service is, in principle read-only for the most part)
<simpson>
dminuoso: Databases have state and versioning, and are expected to mutate. It's like managing active yeast or other cultures; live ingredients need special handling.
<mkaito>
nix doesn't really do state
<mkaito>
and a database is basically pure state
<mkaito>
you can manage the server config
<mkaito>
but the database itself, like roles and such, you'll be doing yourself.
arcushing[m] has quit [Quit: Bridge terminating on SIGTERM]
aduialthin[m] has quit [Quit: Bridge terminating on SIGTERM]
mt[m] has quit [Quit: Bridge terminating on SIGTERM]
lytedev[m] has quit [Quit: Bridge terminating on SIGTERM]
Ericson2314 has quit [Quit: Bridge terminating on SIGTERM]
us3r[m] has quit [Quit: Bridge terminating on SIGTERM]
daemon1024 has quit [Quit: Bridge terminating on SIGTERM]
anderscs[m] has quit [Quit: Bridge terminating on SIGTERM]
logan[m]1 has quit [Quit: Bridge terminating on SIGTERM]
aripiprazole[m] has quit [Quit: Bridge terminating on SIGTERM]
Photonsphere[m] has quit [Quit: Bridge terminating on SIGTERM]
mkg20001 has quit [Quit: Bridge terminating on SIGTERM]
TheSirC[m] has quit [Quit: Bridge terminating on SIGTERM]
drozdziak1 has quit [Quit: Bridge terminating on SIGTERM]
Skallwar has quit [Quit: Bridge terminating on SIGTERM]
pinage404[m] has quit [Quit: Bridge terminating on SIGTERM]
bk1603[m] has quit [Quit: Bridge terminating on SIGTERM]
anthony[m]1 has quit [Quit: Bridge terminating on SIGTERM]
fgaz has quit [Quit: Bridge terminating on SIGTERM]
unclechu has quit [Quit: Bridge terminating on SIGTERM]
shu9 has quit [Quit: Bridge terminating on SIGTERM]
mindtree[m] has quit [Quit: Bridge terminating on SIGTERM]
worldofpeace has quit [Quit: Bridge terminating on SIGTERM]
rycee has quit [Quit: Bridge terminating on SIGTERM]
ejpcmac has quit [Quit: Bridge terminating on SIGTERM]
stites[m] has quit [Quit: Bridge terminating on SIGTERM]
axxoldaccount[m] has quit [Quit: Bridge terminating on SIGTERM]
emily has quit [Quit: Bridge terminating on SIGTERM]
psibi[m] has quit [Quit: Bridge terminating on SIGTERM]
Dandellion has quit [Quit: Bridge terminating on SIGTERM]
nilsirl[m] has quit [Quit: Bridge terminating on SIGTERM]
lambdaclan has quit [Quit: Bridge terminating on SIGTERM]
Piece_Maker has quit [Quit: Bridge terminating on SIGTERM]
rschulman has quit [Quit: Bridge terminating on SIGTERM]
georgyo[m] has quit [Quit: Bridge terminating on SIGTERM]
hsiktas[m] has quit [Quit: Bridge terminating on SIGTERM]
humancalico[m] has quit [Quit: Bridge terminating on SIGTERM]
ncm[m] has quit [Quit: Bridge terminating on SIGTERM]
jschievink has quit [Quit: Bridge terminating on SIGTERM]
betrion[m] has quit [Quit: Bridge terminating on SIGTERM]
DavHau[m] has quit [Quit: Bridge terminating on SIGTERM]
jonathan[m]3 has quit [Quit: Bridge terminating on SIGTERM]
ydlr[m] has quit [Quit: Bridge terminating on SIGTERM]
regnat has quit [Quit: Bridge terminating on SIGTERM]
davidak[m] has quit [Quit: Bridge terminating on SIGTERM]
pheoxy has quit [Quit: Bridge terminating on SIGTERM]
dxb[m] has quit [Quit: Bridge terminating on SIGTERM]
luc65r[m] has quit [Quit: Bridge terminating on SIGTERM]
knotteye[m] has quit [Quit: Bridge terminating on SIGTERM]
neothefox has quit [Quit: Bridge terminating on SIGTERM]
demostanis[m] has quit [Quit: Bridge terminating on SIGTERM]
jonreeve[m] has quit [Quit: Bridge terminating on SIGTERM]
yeongsheng_tan[m has quit [Quit: Bridge terminating on SIGTERM]
eadwu[m] has quit [Quit: Bridge terminating on SIGTERM]
marco[m]3 has quit [Quit: Bridge terminating on SIGTERM]
andreas[m] has quit [Quit: Bridge terminating on SIGTERM]
qbit[m] has quit [Quit: Bridge terminating on SIGTERM]
Spockz has quit [Quit: Bridge terminating on SIGTERM]
Sabbat[m] has quit [Quit: Bridge terminating on SIGTERM]
jdks[m] has quit [Quit: Bridge terminating on SIGTERM]
jgart[m] has quit [Quit: Bridge terminating on SIGTERM]
M0[m] has quit [Quit: Bridge terminating on SIGTERM]
IslandUsurper[m] has quit [Quit: Bridge terminating on SIGTERM]
McSinyx[m] has quit [Quit: Bridge terminating on SIGTERM]
ongy[m] has quit [Quit: Bridge terminating on SIGTERM]
alienpirate5 has quit [Quit: Bridge terminating on SIGTERM]
peeterss[m] has quit [Quit: Bridge terminating on SIGTERM]
bielgeets[m] has quit [Quit: Bridge terminating on SIGTERM]
HumanJohn[m] has quit [Quit: Bridge terminating on SIGTERM]
Ox4A6F has quit [Quit: Bridge terminating on SIGTERM]
nicolas[m] has quit [Quit: Bridge terminating on SIGTERM]
eddyb has quit [Quit: Bridge terminating on SIGTERM]
nasirhm has quit [Quit: Bridge terminating on SIGTERM]
Valodim[m] has quit [Quit: Bridge terminating on SIGTERM]
leons has quit [Quit: Bridge terminating on SIGTERM]
FreeVariable has quit [Quit: Bridge terminating on SIGTERM]
hiroshi[m] has quit [Quit: Bridge terminating on SIGTERM]
doronbehar has quit [Quit: Bridge terminating on SIGTERM]
annemarie[m]1 has quit [Quit: Bridge terminating on SIGTERM]
aterius has quit [Quit: Bridge terminating on SIGTERM]
l-as has quit [Quit: Bridge terminating on SIGTERM]
lasadd[m] has quit [Quit: Bridge terminating on SIGTERM]
reyman[m] has quit [Quit: Bridge terminating on SIGTERM]
Ke has quit [Quit: Bridge terminating on SIGTERM]
tristan[m] has quit [Quit: Bridge terminating on SIGTERM]
faya01[m] has quit [Quit: Bridge terminating on SIGTERM]
boogiewoogie[m] has quit [Quit: Bridge terminating on SIGTERM]
bohan[m] has quit [Quit: Bridge terminating on SIGTERM]
krauser-tan[m] has quit [Quit: Bridge terminating on SIGTERM]
awwayaya24hours[ has quit [Quit: Bridge terminating on SIGTERM]
peterhoeg has quit [Quit: Bridge terminating on SIGTERM]
stuebinm[m] has quit [Quit: Bridge terminating on SIGTERM]
omasanori[m] has quit [Quit: Bridge terminating on SIGTERM]
ThornAvery[m] has quit [Quit: Bridge terminating on SIGTERM]
blindidiotgod has quit [Quit: Bridge terminating on SIGTERM]
princemachiavell has quit [Quit: Bridge terminating on SIGTERM]
q3k[m] has quit [Quit: Bridge terminating on SIGTERM]
txt-file has quit [Quit: Bridge terminating on SIGTERM]
narice[m] has quit [Quit: Bridge terminating on SIGTERM]
Fafnir[m] has quit [Quit: Bridge terminating on SIGTERM]
l33[m] has quit [Quit: Bridge terminating on SIGTERM]
miangraham[m] has quit [Quit: Bridge terminating on SIGTERM]
notgne2[m] has quit [Quit: Bridge terminating on SIGTERM]
JaakkoLuttinen[m has quit [Quit: Bridge terminating on SIGTERM]
felix[m]2 has quit [Quit: Bridge terminating on SIGTERM]
wjjunyor[m] has quit [Quit: Bridge terminating on SIGTERM]
ngerstle[m] has quit [Quit: Bridge terminating on SIGTERM]
dbg[m] has quit [Quit: Bridge terminating on SIGTERM]
iH8c0ff33-matrix has quit [Quit: Bridge terminating on SIGTERM]
aanderse has quit [Quit: Bridge terminating on SIGTERM]
felschr[m] has quit [Quit: Bridge terminating on SIGTERM]
test21562[m] has quit [Quit: Bridge terminating on SIGTERM]
domenkozar[m] has quit [Quit: Bridge terminating on SIGTERM]
SplitFire[m] has quit [Quit: Bridge terminating on SIGTERM]
bbigras has quit [Quit: Bridge terminating on SIGTERM]
BJTH[m] has quit [Quit: Bridge terminating on SIGTERM]
jimkins[m] has quit [Quit: Bridge terminating on SIGTERM]
gcher has quit [Quit: Bridge terminating on SIGTERM]
Alex[m] has quit [Quit: Bridge terminating on SIGTERM]
wenngle[m] has quit [Quit: Bridge terminating on SIGTERM]
srenatus[m] has quit [Quit: Bridge terminating on SIGTERM]
superherointj[m] has quit [Quit: Bridge terminating on SIGTERM]
rednaZ[m] has quit [Quit: Bridge terminating on SIGTERM]
goodnight_noom[m has quit [Quit: Bridge terminating on SIGTERM]
ZerataX has quit [Quit: Bridge terminating on SIGTERM]
cyberwolf[m] has quit [Quit: Bridge terminating on SIGTERM]
sectrect[m] has quit [Quit: Bridge terminating on SIGTERM]
srid has quit [Quit: Bridge terminating on SIGTERM]
scheibenkleister has quit [Quit: Bridge terminating on SIGTERM]
DaExodia[m] has quit [Quit: Bridge terminating on SIGTERM]
tyrion-mx has quit [Quit: Bridge terminating on SIGTERM]
porphyrogenetos has quit [Quit: Bridge terminating on SIGTERM]
bourbon has quit [Quit: Bridge terminating on SIGTERM]
swanson123[m] has quit [Quit: Bridge terminating on SIGTERM]
treed[m] has quit [Quit: Bridge terminating on SIGTERM]
jlv[m] has quit [Quit: Bridge terminating on SIGTERM]
mindset_78[m] has quit [Quit: Bridge terminating on SIGTERM]
foxxxx[m] has quit [Quit: Bridge terminating on SIGTERM]
as3ii[m] has quit [Quit: Bridge terminating on SIGTERM]
manveru[m] has quit [Quit: Bridge terminating on SIGTERM]
inquisitiv3 has quit [Quit: Bridge terminating on SIGTERM]
matthewkenigsber has quit [Quit: Bridge terminating on SIGTERM]
badmutex[m] has quit [Quit: Bridge terminating on SIGTERM]
yoctocell[m] has quit [Quit: Bridge terminating on SIGTERM]
dxu has quit [Quit: Bridge terminating on SIGTERM]
coloneljohnby[m] has quit [Quit: Bridge terminating on SIGTERM]
sum1[m] has quit [Quit: Bridge terminating on SIGTERM]
flip[m] has quit [Quit: Bridge terminating on SIGTERM]
spacetato has quit [Quit: Bridge terminating on SIGTERM]
lukebfox[m] has quit [Quit: Bridge terminating on SIGTERM]
siraben has quit [Quit: Bridge terminating on SIGTERM]
denbrahe[m] has quit [Quit: Bridge terminating on SIGTERM]
oyren[m] has quit [Quit: Bridge terminating on SIGTERM]
DenisLehmann[m] has quit [Quit: Bridge terminating on SIGTERM]
jojosch[m] has quit [Quit: Bridge terminating on SIGTERM]
lg0[m] has quit [Quit: Bridge terminating on SIGTERM]
ilya-fedin has quit [Quit: Bridge terminating on SIGTERM]
ThaEwat has quit [Quit: Bridge terminating on SIGTERM]
das-g[m] has quit [Quit: Bridge terminating on SIGTERM]
hazel[m] has quit [Quit: Bridge terminating on SIGTERM]
callahad[m] has quit [Quit: Bridge terminating on SIGTERM]
ma27[m] has quit [Quit: Bridge terminating on SIGTERM]
felixfoertsch has quit [Quit: Bridge terminating on SIGTERM]
pachumicchu has quit [Quit: Bridge terminating on SIGTERM]
ipv6[m] has quit [Quit: Bridge terminating on SIGTERM]
mcbits[m] has quit [Quit: Bridge terminating on SIGTERM]
zane has quit [Quit: Bridge terminating on SIGTERM]
betux[m] has quit [Quit: Bridge terminating on SIGTERM]
MyriaCore[m] has quit [Quit: Bridge terminating on SIGTERM]
tobim[m] has quit [Quit: Bridge terminating on SIGTERM]
manuform[m] has quit [Quit: Bridge terminating on SIGTERM]
danderson[m] has quit [Quit: Bridge terminating on SIGTERM]
alan[m]1 has quit [Quit: Bridge terminating on SIGTERM]
cirno[m] has quit [Quit: Bridge terminating on SIGTERM]
sarcasticdream[m has quit [Quit: Bridge terminating on SIGTERM]
Notkea[m] has quit [Quit: Bridge terminating on SIGTERM]
ylsdfh27[m] has quit [Quit: Bridge terminating on SIGTERM]
Criena has quit [Quit: Bridge terminating on SIGTERM]
JayQuerie[m] has quit [Quit: Bridge terminating on SIGTERM]
grin[m] has quit [Quit: Bridge terminating on SIGTERM]
gh0la[m] has quit [Quit: Bridge terminating on SIGTERM]
templollete[m] has quit [Quit: Bridge terminating on SIGTERM]
wnklmnn[m] has quit [Quit: Bridge terminating on SIGTERM]
Hm7000 has quit [Quit: Bridge terminating on SIGTERM]
kraem has quit [Quit: Bridge terminating on SIGTERM]
puzzlewolf has quit [Quit: Bridge terminating on SIGTERM]
SpaghettiCthulhu has quit [Quit: Bridge terminating on SIGTERM]
lovesegfault[m] has quit [Quit: Bridge terminating on SIGTERM]
keithy[m] has quit [Quit: Bridge terminating on SIGTERM]
Jezhehim[m] has quit [Quit: Bridge terminating on SIGTERM]
stites[m]1 has quit [Quit: Bridge terminating on SIGTERM]
goibhniu has quit [Quit: Bridge terminating on SIGTERM]
maralorn has quit [Quit: Bridge terminating on SIGTERM]
keash[m] has quit [Quit: Bridge terminating on SIGTERM]
tsrt^ has quit [Max SendQ exceeded]
tet[m] has quit [Quit: Bridge terminating on SIGTERM]
piotrszegda[m] has quit [Quit: Bridge terminating on SIGTERM]
govanify has quit [Quit: Bridge terminating on SIGTERM]
rnhmjoj has quit [Quit: Bridge terminating on SIGTERM]
sethmdoty[m] has quit [Quit: Bridge terminating on SIGTERM]
Layerex[m] has quit [Quit: Bridge terminating on SIGTERM]
musicmatze has quit [Quit: Bridge terminating on SIGTERM]
thefloweringash has quit [Quit: Bridge terminating on SIGTERM]
Yakulu[m] has quit [Quit: Bridge terminating on SIGTERM]
hpfr has quit [Quit: Bridge terminating on SIGTERM]
NobbZ[m] has quit [Quit: Bridge terminating on SIGTERM]
mica[m] has quit [Quit: Bridge terminating on SIGTERM]
philipp[m] has quit [Quit: Bridge terminating on SIGTERM]
axx has quit [Quit: Bridge terminating on SIGTERM]
theduke has quit [Quit: Bridge terminating on SIGTERM]
ryantm has quit [Quit: Bridge terminating on SIGTERM]
timokau[m] has quit [Quit: Bridge terminating on SIGTERM]
alexarice[m] has quit [Quit: Bridge terminating on SIGTERM]
nikola[m] has quit [Quit: Bridge terminating on SIGTERM]
alexfmpe has quit [Quit: Bridge terminating on SIGTERM]
DamienCassou has quit [Quit: Bridge terminating on SIGTERM]
hameerabbasi[m] has quit [Quit: Bridge terminating on SIGTERM]
artturin has quit [Quit: Bridge terminating on SIGTERM]
Minijackson has quit [Quit: Bridge terminating on SIGTERM]
michaelpj has quit [Quit: Bridge terminating on SIGTERM]
roberth has quit [Quit: Bridge terminating on SIGTERM]
jdnixx[m] has quit [Quit: Bridge terminating on SIGTERM]
balsoft has quit [Quit: Bridge terminating on SIGTERM]
srasu[m] has quit [Quit: Bridge terminating on SIGTERM]
leonardp has quit [Quit: Bridge terminating on SIGTERM]
jtojnar has quit [Quit: Bridge terminating on SIGTERM]
TekShifter[m] has quit [Quit: Bridge terminating on SIGTERM]
colemickens has quit [Quit: Bridge terminating on SIGTERM]
slabity has quit [Quit: Bridge terminating on SIGTERM]
DarkTrancer[m] has quit [Quit: Bridge terminating on SIGTERM]
tlater[m] has quit [Quit: Bridge terminating on SIGTERM]
jbal[m] has quit [Quit: Bridge terminating on SIGTERM]
Synth[m] has quit [Quit: Bridge terminating on SIGTERM]
tmeres[m] has quit [Quit: Bridge terminating on SIGTERM]
jakeisnt[m] has quit [Quit: Bridge terminating on SIGTERM]
wak-work has quit [Quit: Bridge terminating on SIGTERM]
crazazy[m] has quit [Quit: Bridge terminating on SIGTERM]
danielrf[m] has quit [Quit: Bridge terminating on SIGTERM]
winsome[m] has quit [Quit: Bridge terminating on SIGTERM]
fatjedi[m] has quit [Quit: Bridge terminating on SIGTERM]
kaziemazie[m] has quit [Quit: Bridge terminating on SIGTERM]
duktus[m] has quit [Quit: Bridge terminating on SIGTERM]
chreekat[m] has quit [Quit: Bridge terminating on SIGTERM]
ptotter[m] has quit [Quit: Bridge terminating on SIGTERM]
nikivi[m] has quit [Quit: Bridge terminating on SIGTERM]
madnat has quit [Quit: Bridge terminating on SIGTERM]
yurb has quit [Quit: Bridge terminating on SIGTERM]
sshow[m] has quit [Quit: Bridge terminating on SIGTERM]
esclear_m has quit [Quit: Bridge terminating on SIGTERM]
zuh0 has quit [Quit: Bridge terminating on SIGTERM]
aniketd[m] has quit [Quit: Bridge terminating on SIGTERM]
Jake[m] has quit [Quit: Bridge terminating on SIGTERM]
JankLoogi has quit [Quit: Bridge terminating on SIGTERM]
kevin[m]1 has quit [Quit: Bridge terminating on SIGTERM]
quidome[m] has quit [Quit: Bridge terminating on SIGTERM]
nh2[m] has quit [Quit: Bridge terminating on SIGTERM]
CRTified[m] has quit [Quit: Bridge terminating on SIGTERM]
neonfuz2 has quit [Quit: Bridge terminating on SIGTERM]
aquarial has quit [Quit: Bridge terminating on SIGTERM]
punsha[m] has quit [Quit: Bridge terminating on SIGTERM]
dbirks[m] has quit [Quit: Bridge terminating on SIGTERM]
witchof0x20[m] has quit [Quit: Bridge terminating on SIGTERM]
nnmmp[m] has quit [Quit: Bridge terminating on SIGTERM]
juanito-terco[m] has quit [Quit: Bridge terminating on SIGTERM]
znello[m] has quit [Quit: Bridge terminating on SIGTERM]
pablo1107[m] has quit [Quit: Bridge terminating on SIGTERM]
zrsk[m] has quit [Quit: Bridge terminating on SIGTERM]
Sajesajama[m] has quit [Quit: Bridge terminating on SIGTERM]
mitchty[m] has quit [Quit: Bridge terminating on SIGTERM]
TimePath has quit [Quit: Bridge terminating on SIGTERM]
kidn[m] has quit [Quit: Bridge terminating on SIGTERM]
<mkaito>
dminuoso: I don't know the particulars of your deployment, but what kind of network outage are you expecting? can't you put your servers on a resilient private network amongs themselves, like EC2 does?
tsrt^ has joined #nixos
<mkaito>
but for read-only use with a single client you can tune postgres to be pretty compact
<mkaito>
if the service is mostly read-only wouldn't a strong caching strategy be better than spamming servers?
<dminuoso>
mkaito: So we're an ISP. This deployment will drive part of our mission critical infrastructure. For that reason, some of these will run on dedicated bare metal nodes directly connected to the hardware components.
Mateon1 has quit [Remote host closed the connection]
<dminuoso>
Access to our hypervisor would traverse our entire core network, firewall, hypervisor rules, and then enter the VM world. Lots of moving parts that could fail
<mkaito>
sounds fun, but that's definitely beyond IRC "advice" lol
<simpson>
nixops itself might not be well-suited to what you want, at least not without a lot of hacking on the bare-metal backend.
dstzd has joined #nixos
joesventek has joined #nixos
nikivi has joined #nixos
<simpson>
But the core concept, which is really just nix-copy-closure and activation over SSH, is definitely reusable.
<mkaito>
tbh if you're just using nixops to push to non-cloud servers, you might be better off just using `nixos-rebuild --target-host foo --build-host localhost`
<mkaito>
I know this sounds like a shill, but check out deploy-rs. it was written exactly for that.
<dminuoso>
The reason we're thinking of nixops, is because there's shared configuration across these things. We have a central writable master node (which you'd dispatch write requests too, but these are allowed to fail in our architecture), so feeding the configuration of that master node to all shards is needed.
<dminuoso>
and equivalently we need to set up streaming replication on all
<Raito_Bezarius>
This part of NixOps is just well done Nix evaluation with shared stuff
<Raito_Bezarius>
You could perform it at an early step, output the Nix config and then copy it over your nodes
<mkaito>
I would argue that is not part of nixops at all and part of nix itself in fact
<Raito_Bezarius>
NixOps is just some Python app which subprocess call Nix and use plugins to manipulate the real world resources into reality and push Nix(OS) over those
<mkaito>
exactly
<mkaito>
if you don't need the part where it interacts with botolib, you probably want to skip nixops. and if you do, I suggest terraform instead.
<simpson>
dminuoso: You're going to have to have a communion with your DBAs to figure out exactly how they want to do stuff like DDL and schemata updates. It'd be nice to use Nix to manage as much of that as possible, but you might not be able to avoid a hybrid approach.
<dminuoso>
simpson: There is no DBA. Schemata updates happen inside the haskell application on startup.
<Raito_Bezarius>
Alternatives are disnix, sops-nix, and more, which have other usecases
<mkaito>
downside to disnix is that the group of people that actually understand it properly include the author and nobody else
Keij0 has joined #nixos
<simpson>
dminuoso: Mildly horrifying, but might facilitate the transition, since you don't have to have Nix manage it for you.
<Keij0>
Hello, I have problems with OpenVPN on NixOS, it doesn't resolve internal DNS at all
<dminuoso>
simpson: Very broadly, my haskell program contains a `migs :: [Migration]` where `data Migration = Migration { migId :: Integer, migQuery :: Connection -> IO () }`, and we keep the current migration id in the database. We check for monotonicity of the migration id on startup, and if it's monotone we just execute remaining migrations.
<dminuoso>
If not, we just fail
joesventek has quit [Quit: Quit]
<Keij0>
VPN's DNS
<mkaito>
I do something similar with rust. Migrations embedded in the binary.
nikivi has joined #nixos
<Keij0>
also I can't estabilish connection with NetworkManager
dstzd has quit [Read error: Connection reset by peer]
<mkaito>
I see. I don't use openvpn or networkmanager, so can't help with that.
dstzd has joined #nixos
<raghavsood>
One option for migrations that are not handled by the program is to have a oneshot systemd unit that checks the db schema and applies migrations if it differs from what is expected
<raghavsood>
That is what I do for some internal packages we use, works alright
joesventek has joined #nixos
<mkaito>
or just run a check in a preStart
<Keij0>
Meh, I thought there's at least one person who uses OpenVPN on here :P
<mkaito>
we used to
<raghavsood>
Wireguard is the future
<mkaito>
yeah, we switched to wireguard and the skies opened and the angels sang
<mkaito>
screw openvpn lol
<Keij0>
.-.
joesventek has quit [Client Quit]
<Keij0>
I know it is a CPU hog but idk if pfSense even supports it
<Keij0>
i mean openvpn is a cpu hog
<dminuoso>
I guess part of why Im keen to use nixops, is because I like the idea of using nixos modules to fully describe my servers.
<mkaito>
I don't know about cpu, but I was mostly bothered by the latency
<dminuoso>
And its within my comfort zone
<mkaito>
dminuoso: that's nix, not nixops.
<Keij0>
but idk if pfSense supports wireguard since it's BSD-based
joesventek has joined #nixos
dstzd has quit [Client Quit]
<Keij0>
is there a limit to how many nameservers I can configure in network.nameservers?
<dminuoso>
mkaito: I have this distinct feeling that I misunderstand what you folks are suggesting.
<srhb>
Keij0: I do use openvpn on NixOS. What's the problem?
nikivi has quit [Client Quit]
<vuko>
Keij0: wireguard works on some BSD's
<Raito_Bezarius>
dminuoso: it's more NixOS than NixOps, NixOps is just a thin wrapper that sends stuff through SSH and nix copy closure
<mkaito>
dminuoso: nixops is literally just a tool that takes your nix and puts it on the server (plus cloud resources, but you don't care about that)
<Raito_Bezarius>
To be more exact, you can use NixOps, but if you have very special requirements, it's better to go for a lower level
<mkaito>
the part that allows you to describe your system config with nix is called NixOS.
<Keij0>
srhb: NetworkManager times out the connection when I try using it, sudo openvpn /path/to/config works, but DNS doesn't work with that
<Keij0>
when I put configs inside configuration.nix it connects, but DNS doesn't work as well
<mkaito>
you can use good old nix config and push it to the server with `nixos-rebuild --target-host target.host.com`
<srhb>
Keij0: Does your ovpn file know how to update your nameservers?
<srhb>
Keij0: Like, it might be trying to use resolvconf from some non-NixOS path
<srhb>
Keij0: Look for the up/down scripts.
<Raito_Bezarius>
But also, if you want to share configuration you can do
nikivi has joined #nixos
<Raito_Bezarius>
You only need to write NixOS modules to factor out your shared things
dstzd has joined #nixos
<Raito_Bezarius>
And you can specialize it in the configuration.nix in the top-level
<Keij0>
it's from work so idk if I can share it publicly
<Raito_Bezarius>
dminuoso: But anyway, give a try to NixOps to confirm if it's what you need
<dminuoso>
mkaito: Well sure I get that, but nixops also gives me stuff like mutually recursive configuration.nix - sure I can probably build that myself somehow
<Raito_Bezarius>
Also, there is #nixops if you need more specific help
<Keij0>
It doesn't look like there's anything relating to nameservers in that config
<Raito_Bezarius>
dminuoso: It's not absolutely mutually recursive iirc
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6my
<thibm>
my DNS is defined in the servers configuration itself, and some of the servers gather all of them and expose them with a DNS server
<mkaito>
I've been exploring baking some of that into the config itself and using it to generate the terraform inputs
<mkaito>
mostly in my own time, in order to generate SRV records for minecraft instances :D
<Raito_Bezarius>
:'D
<mkaito>
you know, _important_ things :P
amir has quit [Quit: Have to disappear. Read you later guys!]
<sphalerite>
mkaito: minecraft supports SRV? Nice
<mkaito>
aye
amir has joined #nixos
sky92 has joined #nixos
<sky92>
hello!
sky92 has quit [Remote host closed the connection]
<dminuoso>
mkaito: Interesting, presumably all of this is deployed with `nixos-rebuild switch --flake '.#'`?
<mkaito>
no, it uses the deploy-rs thing I linked earlier. deployment is defined in the flake as a deploy output.
<mkaito>
but it basically just builds the system closure, nix-copy-closure and activate over ssh, along with some safety checks and automatic rollback.
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6O1
sangoma has quit [Ping timeout: 265 seconds]
<mkaito>
and if you happen to figure out how to package it, I'll be happy to help you write a module, because the day we can move away from BK will be a good day :P
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk63o
<mkaito>
deploy-rs itself hasn't been giving us any trouble, courtesy of Rust being pretty amazing. flakes on the other hand have been giving us some troubles here and there, and the odd day or two debugging C++ has happened.
<dminuoso>
mkaito: So generally what I hear is: "Well if you want to use nix here, dont use nixops. Instead, manually roll your own, use unreleased features or this "this is our edge software", oh and ideally just use nixpkgs master"
<FRidh>
yes, there are indeed some bugs in unstable nix
<srhb>
Keij0: Sorry, work beckoned. It's up to you really. I find that the nicest openvpn config sets specific search domains based on the server config, if the server is indeed pushing domain search and/or routes
<mkaito>
my current pet peeve is that I can't reliably get a debug build of nix on my archlinux machine, and I can't repro half my problems on my nixos server lol
<dminuoso>
It adds more to the "You need to be a nix expert to understand whats going on in this deployment"
<dminuoso>
Reducing the bus factor to 1 or 2 quickly
<srhb>
Keij0: if you're just sending all DNS reqs over the tunnel, nameservers will work just fine.
<mkaito>
we have an entire team just to make nix viable at work. there's a lot of elbow grease involved tbh
<srhb>
Keij0: (At least, until you have the same domain resolving differently outside/inside the VPN)
<dminuoso>
Which I guess is why I instinctively thought of nixops. It's released software, it works, it's stable and it's documented.
<aleator>
Hello! A new Nixos user here. Could anyone point me what to read when encountering '/usr/share/zoneinfo/Africa/Abidjan: File not Found' type of an error?
<mkaito>
nixops is closer to abandoned than stable tbh
<dminuoso>
Isn't nixops 2.0 on the horizon?
<mkaito>
been for a while
<adisbladis>
Tbh nixops 2.0 is pretty close
<mkaito>
nice
<srhb>
aleator: Well, it depends where you're seeing it. It could be an application that isn't patched to find the right path for zoneinfo, for instance.
<dminuoso>
mkaito: In all of this, I havent seen convincing arguments why "nixops" should be avoided here.
thblt has left #nixos ["ERC (IRC client for Emacs 27.1)"]
<dminuoso>
The one annoyance we have, is that it requires ssh password access, but that's already solved in nixops 2.0
<srhb>
nixops is fine, at least until you need to share deployment state (though that may be better in nixops 2.0, I don't know) -- otherwise I prefer the stateless alternative.
<srhb>
Eg. pure nix with some shallow wrapper.
<simpson>
The only arguments against nixops are from experience with nixops, really; it's not perfect, and it shows its issues easily.
<adisbladis>
srhb: The big 2.0 blocker atm is exactly that, dealing with remote state (or stateless deploys)
<dminuoso>
srhb: Just for clarification, when you say `pure nix` you mean `nix-*` and `nixos-*` tools?
<mkaito>
the nixops state file contains SSH keys with root access to all your servers, and is a local binary file that is hard to share. that alone killed nixops for us at work.
<adisbladis>
mkaito: That's also adressed in 2.0. It's entirely possible to disable key mgmt.
<mkaito>
nice, that's one big problem solved
sangoma has joined #nixos
<dminuoso>
mkaito: The local binary file is easily addressed by demanding you use its surrounding shell.nix that provides the binary. And the permissions are bad, ubt solved in 2.0
<aleator>
srhb: In this case, if patching the app is not possible, what other options would there be?
<etu>
adisbladis: Go release 2.0 ;)
<mkaito>
when I say local binary file, I don't mean something you can provide in nix-shell, but state you have to manage yourself in order to keep nixops working.
<dminuoso>
(in 2.0 instead you require a wheel without password user on the target host)
<adisbladis>
etu: :)
<dminuoso>
mkaito: What state is that?
<mkaito>
nixops (currently) keeps track of everything. what's deployed and where and SSH keys too.
<srhb>
aleator: My gut reaction is: I'm sure it's possible to patch the app :)
<mkaito>
this state makes nixops pretty fast compared to tools like terraform or (god forbid) ansible, but it's also state you need to drag around with you
<mkaito>
anyone that wants to run nixops needs this state
<mkaito>
2.0 using the user's SSH key instead of generating new ones and putting them in state is definitely a huge step forward
<mkaito>
because otherwise the state is the most sensitive data known to man and a royal pain in the bum :P
<dminuoso>
mkaito: In this design space, what other alternatives are there?
<srhb>
adisbladis: Cool, thanks for clarifying :)
<dminuoso>
Guess there's also morph
<mkaito>
there's morph, there's just plain nixos-rebuild, and there's deploy-rs and probably two dozen other things
thc202 has joined #nixos
<srhb>
Yeah, everyone and their dog writes a "deploy NixOS" thing at some point in their career. :-P
<mkaito>
the real feature of nixops is mostly allowing you to declare your cloud stuff in nix via botolib and others
<srhb>
Pretty sure we have't reached the global maximum yet :)
<mkaito>
javascript has their is-even libraries, nix has their `nixos-rebuild --target-host` wrappers /shrug
<siraben>
How do I symlink `/nix` to a different location?
<dminuoso>
gah, the one thing I know is I dont want to rely on unstable nix just to do anything
<siraben>
It can't be on the root partition on my embedded device because that partition only has 22 MB
<dminuoso>
Or asked differently, when can we expect a nix release with flakes support?
<mkaito>
dminuoso: to be fair, you can just wrap everything in flake-compat and deploy-rs supports build on stable nix actually. you just need nixFlakes to generate a lockfile.
<mkaito>
3.0 I believe
<siraben>
On macOS it's possible to edit `/etc/synthetic.conf` to point to another volume
<mkaito>
soon (tm)
<mkaito>
bear in mind I don't really know anyone that uses nix and nixos for serious business and doesn't also run somewhere close to unstable
<mkaito>
our servers run off a nixpkgs fork that tracks nixos-unstable-small for example
<mkaito>
dminuoso: in the pegasus-infra that I linked you, you'll see that it has unstable nix and deploy available in devShell, and flake-compat wrappers in shelll.nix and default.nix
<dminuoso>
mkaito: I get it, really. A while ago we pondered running our entire network hardware on nixos..
<simpson>
mkaito: Now you do; I ran in production for a couple years with relatively stable nixpkgs, and certainly with stable Nix. I used boring nixops for management.
<dminuoso>
because as it turns out, our core infrastructure runs on almost plain linux..
<simpson>
Flakes seem really cool. But I don't think that they're production-ready yet.
<mkaito>
well, if stable has everything that you need, that is entirely reasonable. we generally find that stable does in fact not do everything we need :P
<typetetris>
sphalerite: Where can I lookup all those virtualisation attributes?
<mkaito>
mind, we run on the bleeding edge of nix because my entire team are nix and nixpkgs contributors and we just fix things ourselves when it blows up for the most part
<srhb>
mkaito: As a data point, we run a large chunk of the danish library infra on NixOS stable.
<dminuoso>
mkaito: Yeah, we dont fit that description.
<dminuoso>
We have, at best, 2 people including me, who were able to dive into nix all the way like that.
<mkaito>
my company is big on nix
<mkaito>
we usually sponsor nixcon and stuff like that
<dminuoso>
And we cant have a bus factor of 2 on this one
<mkaito>
well, once you fix the warts that affect your stuff directly, other people only need to understand a small part of it
<dminuoso>
mkaito: Ohh you're tweag.
<mkaito>
as long as everything is adequately documented
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6nE
<mkaito>
it then grew and evolved and then people started having a problem with bash because I'm the only one at work that doesn't mind complex bash, so it was rewritten in rust :P
<dminuoso>
I mean rust cant be any more difficult than Haskell..
joem86 has joined #nixos
<dminuoso>
So that's not prohibitive to me
FRidh has quit [Ping timeout: 268 seconds]
<mkaito>
I tried to learn Haskell for years, gave up, and currently really like Rust :P
<mkaito>
which is ironic, given that I work at a haskell shop
SanchayanMaity has joined #nixos
<dminuoso>
Let me guess, you're even hired as a haskell developer
<arahael>
I find the most difficult part of haskell is the tooling and build system.
FRidh has joined #nixos
<dminuoso>
Your haskell skills are just... lazy...
<dminuoso>
not developed until demandede.
<arahael>
And rust seems to have pretty good tooling and build systems.
<mkaito>
haskell was in the job requirements when I got hired, yes.
<mkaito>
but I'm technically an SRE systems engineer
<simpson>
Maybe it's mental modelling of the low-level machine, then. (I'm an SRE specializing in writing code, and I find Haskell and Rust roughly similar in complexity; I prefer Haskell of the two because I get to write less code.)
<mkaito>
I find haskell about as readable as perl tbh
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6cf
<mkaito>
my brain just doesn't into haskell
SanchayanMaity has quit [Client Quit]
amir has quit [Quit: Have to disappear. Read you later guys!]
<{^_^}>
[nixpkgs] @stigtsp opened pull request #104663 → perlPackages.AnyEventAIO: init at 1.1 → https://git.io/Jk6cm
amir has joined #nixos
alp has quit [Ping timeout: 272 seconds]
amir has quit [Client Quit]
amir has joined #nixos
<joem86>
Hi folks. I'm on Fedora now, was considering Silverblue for my next OS when I get my new hard drive, but NixOS seems really interesting to me. My question: would NixOS benefit from being run from a BTRFS filesystem with snapshots? Or would that be redundant to what NixOS already provides by design?
<adisbladis>
joem86: Completely redundant
<adisbladis>
(at least on the system level, your application state is another matter)
<joem86>
Right, that makes sense. Most of my development work is done in the user space anyway.
<simpson>
That is, all of the benefits of the Nix store being relatively immutable and durable and isolated can be reaped on a standard ext4 filesystem. We don't require specific filesystem features, AFAIK.
<joem86>
Nice. Is ext4 the recommended filesystem to use for a fresh installation (for the partition where the nix store would live)?
<adisbladis>
We don't really have a "recommended filesystem"
<simpson>
It's the upstream kernel's recommended general-purpose FS.
<joem86>
Sure. I'm just not aware if any newer filesystems are more performant with symlinks compared to ext4. But so far ext4 has been treating me well.
<siraben>
Installing Nix on i686 here, what's happening after "copying Nix to /nix/store......................................" it's very slow and not printing anything
<thibm>
Uniaika: The environment is defined by the Nix files. If you want to change the environment, change the Nix expression (edit the files or override them, or anything). I think all other ways are hacky
<Uniaika>
thibm: second question: are you aware of any ticket / issue / PR that aims to fix the fact that 'nix-env --uninstall' simply does not warn if it fails / cannot operate?
<Uniaika>
oh hi alp, I was just talking about ghc.nix :)
<siraben>
Uniaika: You should edit the derivation itself
<thibm>
Uniaika: I'm not aware. But I'm not sure what would be the best interface. Should `nix-env -e` fails if the package is not already installed? The answer is not obvious
<thibm>
nix-env -i and nix-env -e are meant (IMHO) to be used by humaan, and they give feedback through the text output
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jk6EU
<thibm>
and finally, I guess that it is expected that the user knows that packages are not "installed" when using nix-shell and thus that they cannot be "removed" by `nix-env -e` (but this assumption may be wrong)
<joem86>
So I understand the upgrade process, when I do "nix-channel --add https://nixos.org/channels/nixos-20.09 nixos", that replaces my existing channel named nixos with the one I specified, right?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/Jk6Dx
sss2 has joined #nixos
qwfpmluy has joined #nixos
<qwfpmluy>
how do you add directories to $PATH in nixos? I tried adding them to `pathsToLink` in my `configuration.nix` but they're still not showing up :/
bulbousbullfrog[ has joined #nixos
joem86 has left #nixos ["WeeChat 2.9"]
boxscape has joined #nixos
<lukego>
Hey is anyone good with cargo2nix? I'm trying it out for the first time and it worked really well to generate Cargo.nix but I'm getting a lot of -- Error: Cannot convert data to TOML (Invalid type <class 'NoneType'>)
FRidh has quit [Ping timeout: 256 seconds]
FRidh has joined #nixos
BJTH[m] has left #nixos ["User left"]
<thibm>
qwfpmluy: what are you trying to do?
<qwfpmluy>
thibm: I want my `cabal` bin dir to be visible in $PATH
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.gidgethub: 4.1.1 -> 4.2.0 »: https://git.io/Jk6dY
<supersandro2000>
red[evilred]: and what has that to do with me?
<red[evilred]>
supersandro2000 (IRC): Absilutely nothing - other than seeing you do all this work is making me feel guilty for some of the things that I have in my queue and have not gotten to
<red[evilred]>
<3
<red[evilred]>
Actually - I do have a random workflow question.
rnhmjoj__ has left #nixos [#nixos]
bbarker has quit [Remote host closed the connection]
bbarker has joined #nixos
<red[evilred]>
How do (plural) you... um, y'all know when a PR/Issue has been seen/replied to when your PR queue grows.
<{^_^}>
[nixpkgs] @jonringer merged pull request #102282 → check-esxi-hardware: refactor from python2 to python3 → https://git.io/JTHH0
<red[evilred]>
before I took on a whole lot more stuff I was just managing one or maybe two PRs at a time
<red[evilred]>
now my list is reaching page limits... how do you know when stuff has moved?
<red[evilred]>
(I'm guessing these some some functionality I've not discovered yet)
<qwfpmluy>
thibm: just putting it in environment.systemPackages seems to not be enough because `echo $PATH | grep '.cabal'` shows nothing. I thought of using `pathsToLink`, thinking that it would then symlink the cabal dir into something like `/run/current-system/sw/bin`
<red[evilred]>
thsi is probably a -dev question
<red[evilred]>
I'll ask there
<red[evilred]>
nm
<thibm>
qwfpmluy: pathsToLink is more of an "internal" option. I think you should not use it except if you are doing advanced stuff
bbarker has quit [Ping timeout: 256 seconds]
<thibm>
cabal-install as an "out" output containing bin, etc and share directory. It should work out of the box whit systemPackages
<thibm>
with*; (just to be sure) did you switch to the configuration after adding it to the systemPackages?
fusion809 has joined #nixos
<lordcirth>
I have a nixos-container that I can't destroy because it says it's declarative; but it's nowhere in my config?
<qwfpmluy>
If with switching config you mean running the usual `sudo nixos-rebuild switch`, yes I did that. Sorry I think I worded my question poorly. cabal itself is showing up in $PATH, I can use it without problems. But I want to access binaries that were created by `cabal new-install` and are placed in `~/.cabal/bin`
<thibm>
OK. Well you should simply add ~/.cabal/bin to your PATH somewhere in your shell intialization files
deadpixels has quit [Remote host closed the connection]
<thibm>
Like programs.bash.interactiveShellInit/loginShellInit if you manage it with NixOS, ~/.bashrc ~/.profile if you manage it classically, etc
<{^_^}>
[nixpkgs] @mweinelt opened pull request #104670 → dash: add patch to prevent code execution when noexec is set → https://git.io/Jk6AU
deadpixels has joined #nixos
<evanjs>
ryantm: is there a sort of scheduler for nixpkgs-update/r-ryantm? i.e. "This package _should_ be updated ~ every week", etc.
<qwfpmluy>
thibm: ok, thanks!
<evanjs>
I opened #101506 before I realized hydrus is updated ~weekly, and I don't really feel like making a PR manually every week, but am totally fine installing, testing, and approving, especially if I'm able to do it with the `nix-store`+binary caches the r-ryantm PRs provide
<srhb>
aanderse: Is there a way to utilize include directives in postgres in 20.09? It seems that all options get an equal sign in `settings` -- just checking if I'm missing something.
bbarker has quit [Remote host closed the connection]
bbarker has joined #nixos
cosimone has joined #nixos
<bulbousbullfrog[>
Can some body give me gist on what nix flakes are for? Their benefits. I'm assuming they will be used more in the future. I just started learning nix so forgive my ignorance
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/JkifM
bbarker has quit [Ping timeout: 240 seconds]
<bulbousbullfrog[>
From that wiki > "Each dependency gets then pinned, that is: its commit hash gets automatically stored into a file - named flake.lock - making it easy to, say, upgrade it"
<bulbousbullfrog[>
I don't get why it makes it easier to upgrade. If it means that you have easier access to bleeding edge packages, won't that make things more brittpe,?
<dminuoso>
bulbousbullfrog[: So lets assume you have some nix expression.
Moonspell has quit [Ping timeout: 240 seconds]
<dminuoso>
We have no general mechanism to bump a given dependency
<dminuoso>
The best thing we have is `nix-channel --update`, but that updates an entire channel
<dminuoso>
If you want to manage a singular component, you devolve into manually writing builtins.fetchGit and manually maintaining those arguments
<dminuoso>
With flakes you get to write `nix flake update --update-input foo`
<bulbousbullfrog[>
I see. If flakes really makes that a lot smoother it would be nice
<dminuoso>
flakes puts nix closer to traditional package managers in that you have a more formal way to talk about and manage dependencies
<bulbousbullfrog[>
Only a few days using nix I already don't like fiddling with got repos their SHAs etc
<bulbousbullfrog[>
* Only a few days using nix I already don't like fiddling with git repos and their SHAs etc
<dminuoso>
Also right now, there's no way to pin dependencies
zupo has joined #nixos
<bulbousbullfrog[>
Using it for dependencies could be useful but then it seems like that could give beginners like myself more power to shoot themselves in the foot
<dminuoso>
the opposite
<dminuoso>
Right now, if you generate a nix derivation that has some inputs and ship that, it claims to accept *any* input versions
zupo has quit [Client Quit]
<dminuoso>
like, it's some `{ foo, bar, ... }: ...` this depends on foo and bar. What are foo and bar? We dont even know. What versions must they be? We dont know
<bulbousbullfrog[>
I see
<dminuoso>
So what you usually do is provide some surrounding builder that provides a pinned nixpkgs
<dminuoso>
But nixpkgs might not have the dependencies on the exact versions you want or need, so you then overlay that with manual fetchGit shenanigans
<dminuoso>
If you dont do all of that, you dont have a truly deterministic setup
<{^_^}>
[nixpkgs] @kampka opened pull request #104673 → toybox: Add option to enable shell support → https://git.io/JkiU5
<bulbousbullfrog[>
So somebody with more know how can provide sane inputs as flakes to make sure everything goes as planned
<dminuoso>
(because it would rely on some external ambient state: the nix channel)
<dminuoso>
well, with flakes it's just pinned automatically
<dminuoso>
right now the default is "automatically updates with your ambient nix-channel", with flakes the default is "pinned until you tell me to upgrade"
<bulbousbullfrog[>
That's cool then. Thanks for explaining. The benefits are much clearer to me now
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkikP
<tobiasBora>
For example, if I run "steam-run bash", I get an error "/nix/store/mbrhnpc8a5kh5pyp92hnfbnlyywp5qdx-gnupg-2.2.23/bin/gpg-connect-agent: symbol lookup error: /lib64/libc.so.6: undefined symbol: _dl_fatal_printf, version GLIBC_PRIVATE"
<lordcirth>
tobiasBora, what NixOS version? Did you update, especially glibc?
<tobiasBora>
If inside this shell, I run the Natron portable executable (that was working before), I get an error symbol lookup error: /lib64/libfontconfig.so.1: undefined symbol: FT_Done_MM_Var
<tobiasBora>
lordcirth: I upgraded recently (but not exactly yesterday as far as I remember) to nixos 20.09, commit 58f9c4c7d3a.
<tobiasBora>
lordcirth: so it may upgraded glibc as well. But I tried to reboot, hoping to see some improvements, but the errors is still here
<sss2>
any ideas ?
Moonspell has joined #nixos
ilyaigpetrov has joined #nixos
deadpixels has quit [Ping timeout: 260 seconds]
<lordcirth>
tobiasBora, do you have any overlays or other channels?
<tobiasBora>
sss2: it seems that you upgrade the munkres python library, and for some reasons some tests of this library fail. I can see that 16 days ago, jonringer wrote a fix to this package that was supposed to fix the test phase, so it may be related. Maybe this fix was not backported? Which system are you running?
<tobiasBora>
lordcirth: I don't really have overlays, but I do install some stuff from unstable (emacs) and from 20-03 (sageWithDoc, kdenlive and kmymoney)
philr has quit [Ping timeout: 256 seconds]
acarrico has joined #nixos
<sss2>
is it possible to build this package locally and safe it to store to avoid distributed builds for just this single package ?
<tobiasBora>
lordcirth: is it supposed to be a problem if packages are installed from several channels with different glibc versions?
<lordcirth>
tobiasBora, no, it shouldn't. Nix is very good at that
<tobiasBora>
lordcirth: hum... also, if it helps, I've home-manager running, not sure if it changes something
<evanjs>
ryantm: alright. So I could either do nixpkgs-update myself (which is still an improvement) or just let r-ryantm update it when it does. Which is fine with me as I'm mostly using the program on Windows right now, anyway. Thanks!
<evanjs>
ryantm++
<{^_^}>
ryantm's karma got increased to 27
dtz has quit [Quit: Idle for 30+ days]
srxl has quit [Quit: Idle for 30+ days]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<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
cosimone has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Fuseteam opened pull request #104683 → fix compilation on armhf → https://git.io/Jki8X
cosimone has joined #nixos
jess has joined #nixos
<{^_^}>
[nixpkgs] @hamburger1984 opened pull request #104684 → Upgrade fetch-kde-qt.sh to support current layout of download.kde.org → https://git.io/Jki4D
<Abdullah>
so how do you recommend nixos to new user?
<Abdullah>
not new to linux.
<dbe2>
Is the /nix/var/nix/profiles/system symlink necessary? I'm trying to install a nixos flake into an image, and I figured that I can do "nix copy --to dir flake" then bind-mount sys,proc,proc, chroot into the directory and run the activate script on the nixos derivation to create all the necessary directories... feels like the system profile isn't necessary then?
<dbe2>
Abdullah: what pitch are you interested in? nixos has lots of great feature!
dbe2 is now known as dbe
<Abdullah>
I just want some new distro to try. getting bored in this installation and its almost a year I haven't tried something new
<Abdullah>
I just want an OS for my laptop
<Abdullah>
with ssd.
<Abdullah>
I'd like to know about ZFS if its good for ssds as compare to ext4.
<Abdullah>
pluse I always do some encryption while installing. luks on lvm
<{^_^}>
[nixpkgs] @jonringer pushed commit from @dpaetzel to master « pythonPackages.mesa: unstable-2019-12-09 -> 0.8.7 »: https://git.io/JkiR2
<lordcirth>
Abdullah, ZFS root on SSDs is great. I have root on a raidz of 3 SATA SSDs
<dbe>
Abdullah: nixos is neat. Just try it out and see how it feels. It can do some really cool stuff. If you have time to spend, why not.
<Abdullah>
I love it <3
<dbe>
Abdullah: Personally, I love being able to write flake.nix into a repository, forget about it for a year, and go back to it and just run "nix shell" and have all the development tools I need. It's useful for work.
<Abdullah>
lordcirth: so ZFS is something like to create RAIDs? I have only one SSD and no other hard drive.
bbarker has quit [Remote host closed the connection]
bbarker has joined #nixos
<lordcirth>
Abdullah, ZFS works great on one device too! You still get snapshots, checksumming, and multiple filesystems
<Abdullah>
hmm
<supersandro2000>
dbe: "Enterprise ready forget that there are updates" included
<Abdullah>
supersandro2000: forget that there are updates?
<Abdullah>
what that means?
<lordcirth>
I love being able to install some packages, enable some services, git push from my PC and git pull them on my laptop.
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkiRx
<Abdullah>
and what about packages?
<lordcirth>
Abdullah, like, you can write software on NixOS 19.09, come back to it a year later running NixOS 20.03, and the software you wrote will be automatically built against the exact same libs you did a year ago, unless you choose to update flake.nix
<lordcirth>
(or flake.lock)
<Abdullah>
do I know what's flake.nix?
bbarker has quit [Ping timeout: 240 seconds]
<lordcirth>
Abdullah, flake.nix is the entry point into a "flake" which is a method of describing nix builds. Whether that's building one program, as dbe is probably talking about, or my entire OS in my case.
<Abdullah>
what about its packages?
<lordcirth>
My OS install, config, users, services, packages, etc are all described in one git repo that I can pull onto another NixOS machine and switch to.
<Abdullah>
I mean the package manager?
cosimone has quit [Quit: cosimone]
<lordcirth>
The package manager is Nix, and most of the cool stuff about NixOS is really Nix features.
<{^_^}>
[nix] @edolstra merged pull request #4275 → fetchMercurial: set HGPLAIN when invoking hg → https://git.io/Jkinf
<Abdullah>
I was reading that you can roll back to previous state. I think that needs grub or some other boot manager. I don't use boot manager. I use EFISTUB. so is there some problem?
<lordcirth>
Abdullah, there are similarities, in being able to customize heavily. But I'm not aware of any other distro that has declarative config / package management built in.
<lordcirth>
Abdullah, if you totally break your boot, you might need a boot manager to roll back. But you can roll back from even a very broken system, if you can get a shell.
FRidh has quit [Ping timeout: 240 seconds]
<lordcirth>
I can switch from NixOS 20.03 to 20.09 to unstable and back again without rebooting.
<Abdullah>
I have time to keep working on it even have a YouTube channel to promote nixos if I liked it ;-)
<Abdullah>
impressive
<lordcirth>
(Well, Guix exists, but it's kinda a Nix clone, so...)
mallox has joined #nixos
<cole-h>
s/kinda //
<Abdullah>
I think I tried it guix which has no non-free software? right
mallox has quit [Client Quit]
evanm has quit [Ping timeout: 245 seconds]
pushqrdx has joined #nixos
<{^_^}>
[nixpkgs] @Luflosi opened pull request #104687 → nextpnr: unpin Qt version → https://git.io/JkiEg
<{^_^}>
[nixpkgs] @Luflosi closed pull request #104682 → nextpnr: fix build on Darwin → https://git.io/JkiW5
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<zeorin>
Hey all. I just tried to install the latest Nix on my Ubuntu machine. I did a multi-user install, and the output at the end said it was successful. But I cannot actually access any nix commands
<nschoe>
zeorin: kill your shell and restart a new one.
<nschoe>
You don't have the .sh file sourced.
<zeorin>
Yeah I've done that…
iH8c0ff33 has joined #nixos
<zeorin>
I didn't (and still don't) have /etc/bashrc or /etc/zshrc files
<zeorin>
But /etc/profile.d/nix.sh was created
<nschoe>
zeorin: then source that :)
<zeorin>
Is that not normally already sourced?
<nschoe>
zeorin: and then add 'source /etc/profile.d/nix.sh' in your *sh init file
<{^_^}>
[mobile-nixos] @samueldr pushed 2 commits to master: https://git.io/JkiXF
deadpixels has joined #nixos
<simpson>
oxalica: FWIW all of your PRs look good. It seems like the bureaucracy has failed here.
<nschoe>
zeorin: it's been a while since since I last tried Nix on a non-NixOS system, so I can't remember. In my memories, it did modify your .bashrc or smth. But I'm not sure.
<gustavderdrache>
you may need to force a login shell - try `. ~/.profile` and see if that gets you anything
<{^_^}>
[nixpkgs] @fabianhjr opened pull request #104699 → recoll: refactor and 1.24.5 -> 1.27.12 → https://git.io/Jki1L
<cole-h>
simpson: IMO, not that bureaucracy has failed -- just that very few people are interested in those packages, and even fewer (if any) have w+
<simpson>
cole-h: Then worse: There is no bureaucracy.
<zeorin>
Hmm manually sourcing that works.
<cole-h>
¯\_(ツ)_/¯
<zeorin>
I'm using zsh fwiw
<cole-h>
Not much to be done when you have 2.5k PRs wanting to be merged and only 144 people with w+
<simpson>
(A DMV would not be considered successful if it only speedily processed the claims of popular folks and folks with power.)
<zeorin>
Perhaps that's not sourcing anything in /etc/profile.d
<cole-h>
No, but the DMV also (allegedly) has the manpower of more than 144 people :P
<simpson>
cole-h: Flipping that around, shouldn't it only take an afternoon for each of those maintainers to review and close ~20 PRs? I don't think that it's a question of manpower, exactly.
<cole-h>
Those at the DMV are also being paid to be there and sort through their N guests
<samueldr>
many open PRs are in the non-trivial realm, too
<simpson>
I could have sworn that some folks are making money off of nixpkgs; surely some people are paid to be here~
<cole-h>
It might only take an afternoon, but what does that matter if they can't make time for that afternoon.
<samueldr>
no one is paid _to review PRs_ AFAIK
bbarker has quit [Remote host closed the connection]
<cole-h>
^
<samueldr>
paid adjacent to working with Nixpkgs
berberman has quit [Ping timeout: 264 seconds]
berberman_ has joined #nixos
<samueldr>
but not paid strictly for Nixpkgs' own sake AFAIK
<simpson>
Okay. So, just to double-check that I grok: nixpkgs has a bunch of pending work from the community which isn't getting merged, a bunch of people making money from community contributions but not putting paid time towards gardening, and there's nothing much that can be done about this. Yes?
<cole-h>
I don't think I'd agree with "a bunch of people"
<samueldr>
a few people that improve Nixpkgs for their own, or client's, needs rather than "making money from community contribution" which seems to want to imply malicious intent
<simpson>
Not malicious intent, just a commons (as in the Tragedy of the Commons) which is open for exploitation but Somebody Else's Problem when it needs to be improved or maintained.
<zeorin>
Fixed. Added emulate sh -c 'source /etc/profile' to /etc/zsh/zprofile. Supposedly this is supposed to be already set up by Ubuntu, but in my case it wasn't
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkiMi
<samueldr>
though yeah, it'd really help if we had people who's actual _job_ was to make contributions flow through
<simpson>
samueldr, cole-h: I *can* point out technical barriers, too, FWIW, in oxalica's PRs. For example, one PR seems stalled simply because it does a lot of rebuilds; we should invest effort into tooling which can coalesce rebuilds. (Hoping that somebody can tell me that it already exists.)
<cole-h>
DNE, but has been discussed (e.g. Rust's bors bot was used as inspiration)
<cole-h>
simpson: I don't think the openssl stalled due to its rebuilds -- more because peti doesn't really engage outside of Haskell anymore. A better person to ping would be FRidh, who does most of the staging stuff.
<cole-h>
s/openssl/openssl PR/
zeorin has quit [Ping timeout: 245 seconds]
<ryantm>
@ivan why does that package need the lockfile to be patched?
<ivan>
ryantm: a lot of Rust projects don't have a Cargo.lock in their git and nixpkgs's buildRustPackage needs one
<ryantm>
ah, thanks!
<cole-h>
But why was it moved?
<simpson>
cole-h: Again, just to double-check that I grok: Community contributions need to flow through the appropriate power broker, and the brokers must be appeased by a token appeal of some sort? Like, this happens in Linux as well, but in that tree there are specified people who are *responsible* for those merges; their power is balanced by responsibility.
<simpson>
(I've personally had very silly patches merged into Linux simply because I emailed them to the right person, and they had a script which pulled everything from their email into git. I didn't break anybody, so my work got included.)
deadpixels has quit [Remote host closed the connection]
<samueldr>
personally, there's a lot of PRs which exercise a corner of Nixpkgs I don't know about
<samueldr>
so when I see those, I back away
<samueldr>
e.g. `python3Packages.buildPythonPackage`, I don't know what gotchas to look for, so I don't
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
<supersandro2000>
grahamc[m]: love to spot ofborg issues
<ryantm>
FRidh: how did you decide what to close?
<{^_^}>
[nixpkgs] @jonringer merged pull request #104670 → dash: add patch to prevent code execution when noexec is set → https://git.io/Jk6AU
<{^_^}>
[nixpkgs] @jonringer pushed commit from @mweinelt to master « dash: add patch to prevent code execution when noexec is set »: https://git.io/JkiHd
<supersandro2000>
I already thought about reverting but I wasn't sure and build it locally
<gchristensen>
supersandro2000: heh, what do you mean?
<gchristensen>
100-500 usually not, we can usually do 100 builds quite quickly in hydra
<supersandro2000>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 377
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jonatanb has quit [Remote host closed the connection]
<ryantm>
r-ryantm currently targets staging at 100, should I change that cutoff to something else?
<gchristensen>
this is of course the 10,000 build question that nobody has been willing to put a specific number on :P
<FRidh>
no please keep that
<FRidh>
as soon as there are multiple 100-500 merged, the impact can be quite big
<ryantm>
Probably makes sense for the threshold for regular contributions to be higher than r-ryantm ones.
<FRidh>
I typically merge them in bulk before a new staging-next cycle
<FRidh>
unless someone else of course merged them already
zeorin has joined #nixos
<danderson>
Where does the release notes doc live these days? The one in nixpkgs/doc is really old, afaict
<FRidh>
nixos/doc
<danderson>
aha, thx
<zeorin>
Hello all, I just installed Nix on my ubuntu machine. I have a nixos machine, so I'm a little used to nix already. What's strange to me is that I don't seem to have any channels set up, yet I'm able to install software using nix.
<zeorin>
nixos of course comes with the stable channel already set up
<zeorin>
Should I explicitly set up a channel?
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
<{^_^}>
[nixpkgs] @mweinelt pushed to release-20.09 « dash: add patch to prevent code execution when noexec is set »: https://git.io/Jki7L
zupo has joined #nixos
rprije has joined #nixos
szicari has quit [Ping timeout: 256 seconds]
malook has joined #nixos
cosimone has quit [Remote host closed the connection]
<meatcar>
What's the concensus on using `with` for single-item lists, like `maintainers = with maintainers; [ foo ];` vs `maintainers = [ maintainers.foo ];`? `with` scares me a bit in general, and within a `rec` especially, so I try to avoid it where possible.
lambda-11235 has joined #nixos
<srhb>
meatcar: It's fine either way. And being wary of it is too. As long as you keep its scope very local, it's usually not confusing or dangerous.
<srhb>
zeorin: Likely root has a channel and you're using that without realizing?
<zeorin>
Yeah I just figured that out. Root has a channel `sudo nix-channel --list` complained about nix-channel not being a valid command
<ryantm>
meatcar: I like the former because it makes it easy to add a new maintainer.
<zeorin>
`sudo -i nix-channel --list` informed me that the `nixpkgs` channel is set to unstable
<srhb>
zeorin: That sounds normal on non-NixOS linux.
deadpixels has quit [Remote host closed the connection]
<meatcar>
If only there was a similarly short an concise way to do `builtins.attrValues { inherit (maintainers) foo; };` :)
zupo has quit [Ping timeout: 265 seconds]
szicari has joined #nixos
iH8c0ff33 has quit [Ping timeout: 265 seconds]
zupo has joined #nixos
gthm has quit [Ping timeout: 260 seconds]
Morfio has quit [Quit: This computer has gone to sleep]
<bsima>
I'm trying to compile a package that depends on libfive-guile, but I can't figure out how to get that library in the right context. What am I missing here? Nix code: http://ix.io/2FeM / Error: http://ix.io/2FeN
zeorin has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
vidbina has joined #nixos
szicari has quit [Ping timeout: 264 seconds]
szicari has joined #nixos
alp has quit [Ping timeout: 272 seconds]
alp has joined #nixos
_deepfire has quit [Read error: Connection reset by peer]
<la-jesystani>
hey, im feeling pretty dumb here ive been reading docs all night, could someone please tell me how to use the nixops flake, and tell it which plugins i need?
<{^_^}>
[nixpkgs] @SuperSandro2000 merged pull request #104698 → yed: mark as broken if jre.gtk3 is not available → https://git.io/JkiKM
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkibD
vidbina has quit [Quit: vidbina]
iH8c0ff33 has joined #nixos
vidbina has joined #nixos
vidbina has quit [Client Quit]
mkaito has quit [Quit: WeeChat 2.9-dev]
bbarker has quit [Ping timeout: 256 seconds]
bbarker has joined #nixos
vidbina has joined #nixos
Darkmatter66 has joined #nixos
iH8c0ff33 has quit [Ping timeout: 240 seconds]
lsix has quit [Ping timeout: 264 seconds]
<tokudan>
how do i create zfs datasets during boot before systemd tries to mount local filesystems?
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkiA3
<tokudan>
lordcirth, idea is that when I add a service to a system i just add a dataset to a list and configure it as a filesystem. then nixops deploy --force-reboot and i'm done
Reventlov has quit [Quit: WeeChat 2.3]
lsix has joined #nixos
<lordcirth>
tokudan, so every stateful service will have it's state in a seperate filesystem? Seems kinda overkill but ok
<tokudan>
lordcirth, yeah, and i can just login to the system and do a zfs snapshort -r rpool/data followed by zfs send -r... rpool/data@backup to get a full backup of the system
Reventlov has joined #nixos
Reventlov has joined #nixos
Reventlov has quit [Changing host]
<tokudan>
i've been playing around with local-fs-pre.target, local-fs.target and zfs.target, but none of them seem to help me in any way
<lordcirth>
tokudan, I would just make /var a filesystem
<raboof>
I'm trying to build graphviz in nix-shell, but it seems I'm doing something wrong.
<lordcirth>
Or even noauto and have a service prestart or init service handle it
<tokudan>
hmm... that might work
<raboof>
after entering `nix-shell '<nixpkgs>' -A graphviz` I can `unpackPhase`, but entering the source directory and running `configurePhase` doesn't seem to do anything
domogled has joined #nixos
<raboof>
I guess I need to invoke autotools somehow, but how?
<tokudan>
needs some extra work on some services, but it should work
<JonReed>
Why could possibly something work when it is started inside xsession, but have bugs when it is started via a systemd service? I was trying to start `xbindkeys` as a systemd service in NixOS for hours now and nothing I try works. Eventually, I found this https://logs.nix.samueldr.com/nixos/2017-10-07#611290 comment saying that they also weren't able
<JonReed>
to start it as a systemd service. I copy the exact same command that the service generates in ExecStart and run it in console and it works. But when I start the service with `systemctl --user start xbindkeys` xbindkeys does't work (it has a weird bug where I see it more and more of xbindkeys processes spawning each time I press a key combination
<JonReed>
and after a couple of them it crashes - this doesn't happen anywhere outside systemd).
<JonReed>
It sees the connect "DISPLAY" too, because when I deliberately change that to an invalid one, it refuses to start. So, like what the hell is different? How running a command via `systemd --user start service` is different from running it from console? I'm so confused.
<JonReed>
s/connect/correct
<JonReed>
infinisil: Sorry for the direct ping, but you're like the only guy who on the Internet who I found had the same problem. Were you able to find out why xbindkeys weren't working for you as a systemd service or I should give up and let it forever remain a mystery?
<infinisil>
Oh wow, that was a while ago :o
<{^_^}>
[nixpkgs] @jaredramirez opened pull request #104709 → elmPackages: add elm-graphql and elm-review → https://git.io/Jkijg
cosimone_ has joined #nixos
dxtr has joined #nixos
<JonReed>
infinisil: :D I know. I assume if somebody like you weren't able to solve it in 3 years, then I should give up right now.
cosimone has quit [Ping timeout: 260 seconds]
cosimone_ is now known as cosimone
<infinisil>
I don't think I ever figured that out
boxscape has quit [Quit: Connection closed]
<infinisil>
JonReed: Maybe try comparing `env` within systemd and `env` in a terminal
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkPUL
bbarker has quit [Ping timeout: 246 seconds]
mahogany has quit [Quit: Konversation terminated!]
Ariakenom has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @thomasjm opened pull request #104714 → Extend nix-prefetch-git to support passing tree hashes as "rev" → https://git.io/JkPUb
<bsima>
halfbit: nativeBuildInputs is runtime deps so they would need to be included in the resulting closure, right?
<pushqrdx>
can someone try to run ghidra (the reverse engineering tool) and tell me if it runs correctly?
<pushqrdx>
cause it runs here but no text is visible
<pushqrdx>
show blank background
<cole-h>
bsima: No, nativeBuildInputs are not runtim deps.
<cole-h>
runtime*
<cole-h>
They are just that: build inputs.
<halfbit>
bsima: oh maybe I'm using the wrong thing then, I need to specify certain tools/libs for compile time
<cole-h>
nativeBuildInputs is used for tools like cmake, gnumake, ninja, etc that are ran during build time
<halfbit>
the cross compile docs seemed to make me think nativeBuildInputs was the thing, it also linked to whats probably a relevant issue about wrapping things in callPackage, which I do, but it still doesn't work
<cole-h>
When cross-compiling, you need to use the packages listed in nativeBuildInputs from the targeted system (IIRC)
<cole-h>
Otherwise, it will use the binaries suitable for your current system
selfsymmetric-mu has joined #nixos
<halfbit>
like I have default.nix with { pkgs ? import <nixpkgs> { crossSystem { config = "..."; } }} pkgs.callPackage( {pkgs}: pkgs.stdenv.mkDerivation { nativeBuildInputs = [cmake]; buildInputs = [zlib]; } ) sort of like that
<halfbit>
nix build then compile cmake for whatever the crossSystem config is set to
<halfbit>
armv7-unknown-linux-musleabihf for example
selfsymmetric-mu has quit [Remote host closed the connection]
<halfbit>
I think I need to test some things out in a smaller test project first heh
wfranzini has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/JkPkk
selfsymmetric-mu has joined #nixos
<selfsymmetric-mu>
I installed steam as `pkgs.steam.overrideAttrs (orig: { extraPkgs = with pkgs; [ qt5.qtbase ]; })` but I'm still getting `vrmonitor: error while loading shared libraries: libQt5Core.so.5: cannot open shared object file: No such file or directory`. What am I doing wrong?
<JonReed>
infinisil: Thanks! It was related to environment vars somehow. I figured out that if, instead of starting it directly, I wrap it using `wrapProgram` in a way where all environment variables are unset and then $PATH and $DISPLAY are set back again, it works. Vice versa, taking the command in ExecStart and running it in a `nix-shell --pure` sandbox
<JonReed>
with the $PATH copied from the service's init file that was generated, produced the same error. Yay! Now I can narrow it down even more.