<tobiasBora>
clever: ohh... you mean like binary = "${pkgs.weechat}/bin/weechat --dir <HOME DIR>"? I just don't see how to setup <HOME DIR> with something like pkgs.writeText, my first guess would have been to put it directly to /etc/myweechat_conf/ using environment.etc, but maybe writeText has an alternative for folders?
<clever>
tobiasBora: whatever you put in binary, will get passed to screen, so you could also sneak screen flags in via binary = "-something ${weechat}/bin/weechat"; lol
<tobiasBora>
clever: I see... That's super interesting. And one advantage over environment.etc is that it actually copies when you move derivation from computer from computers also?
<infinisil>
> zfsUnstable.name # pie__
<{^_^}>
"zfs-user-0.8.0"
<pie__>
is there some easy way to restart my network stack
<pie__>
its in some weird half broken state
<clever>
tobiasBora: more that it can work without nixos
<pie__>
or maybe its just firefox
<clever>
tobiasBora: and you can test it with just nix-build alone
<clever>
tobiasBora: the weechat wrapper does similar
Fare has joined #nixos
zeta_0 has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « p11-kit: disable tests on darwin »: https://git.io/fjuTm
iqubic` has left #nixos ["ERC (IRC client for Emacs 26.1)"]
fusion809_ has quit [Remote host closed the connection]
fusion809 has joined #nixos
<tobiasBora>
clever: indeed, but it also uses the '/script load' trick. But I just realized that there is this "WARNING: It is NOT recommended to edit this file by hand [...] Use /set or similar command to change settings in WeeChat." I guess I'll just try to use /set in the init script. But the writeText stuff was very interesting, thanks!
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjuTY
<gyroninja__>
>substitute is implemented using the replace command. Unlike with the sed command, you don’t have to worry about escaping special characters
<gyroninja__>
This is untrue, I had to escape a $ in order for one of my substitutions to work
schjetne has joined #nixos
<gyroninja__>
I checked the derivation and the replacement string properly had a $ in it
<clever>
gyroninja__: its more about escaping it from bash when calling substitute, rather then / and \ and others being special to sed
<pie__>
(it may warrant a skillful clarification)
<freshee>
Hi. I have a small question. I need a file (an ssh key) to have 0600 permissions. I thought of running `chmod 0600 $out/etc/id_rsa` in `installPhase`, but the final permission is 0444. Why would that happen?
<clever>
freshee: all files in /nix/store must be world readable
simon_weber has quit [Quit: Connection closed for inactivity]
tobiasBora3 has quit [Quit: WeeChat 2.4]
zeta_0 has quit [Quit: fixing .emacs]
zeta_0 has joined #nixos
freshee has quit [Quit: Page closed]
drakonis has quit [Ping timeout: 252 seconds]
tobiasBora2 has joined #nixos
drakonis has joined #nixos
<pie__>
infinisil, i dont realllly want to switch to unstable though given that 0.8 has a stable version now (?)
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « qtwebengine: don’t use gn configure »: https://git.io/fjuTQ
<infinisil>
pie__: 0.8 is stable, that's why it has a release number lol
mexisme_ has joined #nixos
<infinisil>
ah
<infinisil>
sorry I misread stable/unstable
schjetne has quit [Ping timeout: 252 seconds]
<infinisil>
pie__: Yeah, it's not optimal, but only when some people have verified that the ZFS release works without problems on NixOS, zfs will be pointed to 0.8
<infinisil>
Until then it's still "unstable" for NixOS
<clever>
and nixos will list all of the failures, rather then just the first one
<pie__>
i will just use the first thing infinisil posted lol
<clever>
pie__: assertions = [ { assertion = pkgs.zfsUnstable.name == "zfs-user-0.8.0"; message = "zfs version is trying to change"; } ];
<pie__>
actually, thats good to know though
<eddyb>
lucus16: since `nix-build` seems to be working now, I'm going with `doCheck = true; installTargets = "fail-on-purpose";`
polman has joined #nixos
<clever>
amd feature@large_dnode disabled local
<clever>
amd feature@edonr disabled local
<clever>
pie__: my desktop has 2 other features not yet enabled
<clever>
The large_dnode feature allows the size of dnodes in a dataset to be set larger than 512B.
drakonis has quit [Ping timeout: 258 seconds]
<clever>
This feature becomes active once a dataset contains an object with a dnode larger than 512B, which occurs as a result of setting the dnodesize dataset property to a value other than legacy.
<clever>
will return to being enabled once all filesystems that have ever contained a dnode larger than 512B are destroyed
<clever>
pie__: so this flag requires changing properties to non-default values, to activate
<clever>
and destroying the entire dataset to de-activate
<pie__>
clever, can you set it back to the default value and try to remove all the bad things
<pie__>
"bad things"
<clever>
pie__: zfs may not be able to track how many "bad things" remain
<clever>
and only knows that the entire dataset has become suspect
drakonis has joined #nixos
<pie__>
scrub?
<clever>
good point, that could maybe update that flag
<clever>
`This feature enables the use of the Edon-R hash algorithm for checksum, including for nopwrite`
<clever>
the other one is just a hashing algo, which can then be used as the checksum for all blocks in the dataset
<clever>
nopwrite is similar but much cheaper then dedup
<clever>
for dedup, you have a massive hash table, to see if a block with hash(content) exists, and where it is
<clever>
but with nopwrite, your overwritting block X, and can just compare the checksum on-disk, to the checksum(newcontent) to see if it actually changed
<clever>
no expensive lookup, no hashtable to power it
Fare has quit [Ping timeout: 272 seconds]
<greenerworld[m]>
should I poke distrowatch for including unstable in releases?
<greenerworld[m]>
should I create an issue about it somewhere? or just do it
<gchristensen>
up to you :)
<ashkitten>
infinisil: fwiw, i've been using zfsUnstable for quite a while (due to a mistake on my part) and had no problems with it even with release candidates
<samueldr>
wth is going on with that logo?
<samueldr>
right shape, old colours
<gchristensen>
mysteries, sa
<infinisil>
ashkitten: Maybe you can comment in #61969
<clever>
[clever@system76:~]$ man /nix/store/pf10i5lvzqgg3b1jy00giz84zv05lplf-zfs-user-0.7.13/share/man/man8/zfs.8.gz | head
<clever>
ZFS(8) System Manager's Manual ZFS(8)
<pie__>
samueldr, im a nano noob
<samueldr>
pie__: don't feel bad, I just like the diff view from [n]vim
<pie__>
i just did diff --color=always <(man /nix/store/msrcfzv0i9k1dgj9dbpqkm9gnks1q7xy-zfs-user-0.8.0/share/man/man8/zfs.8.gz | more) <(man /nix/store/pf10i5lvzqgg3b1jy00giz84zv05lplf-zfs-user-0.7.13/share/man/man8/zfs.8.gz | more )
<samueldr>
funny, more unboldens it
<pie__>
though now im trying to figure out how to get unchanged results to show up too
<pie__>
samueldr, i just figured based on less needing a -R flag for some stuff that htis might work
<pie__>
soooo teeeeeechnically zfs isnt doing full disk encryption
<pie__>
"ZFS will not encrypt metadata related to the pool structure,
<pie__>
itself is encrypted)."
<pie__>
including dataset and snapshot names, dataset hierarchy, properties, file size, file holes, and deduplication tables (though the deduplicated data
domogled has quit [Quit: domogled]
<clever>
pie__: then dont make any datasets called illegal-porn and youll be safe :P
slack1256 has quit [Ping timeout: 252 seconds]
sb0 has quit [Quit: Leaving]
InFlames has joined #nixos
<pie__>
so many caveats
<pie__>
:P
<clever>
simpler to just use luks
<pie__>
to my naive self all that sounds fine but i dont know about the deduplication tables, and deduplication informatiom might leak some information about file data?
reinhardt has joined #nixos
<clever>
pie__: dropbox had a related exploit
<clever>
pie__: originally, you sent the hash of a file to the server, and if dropbox already had a copy, its instantly added to your box
<clever>
skip the upload!
<clever>
then somebody realized, if they try to upload an illegal file, and it doesnt have to upload, somebody is in trouble :P
<pie__>
clever, doh lol, thats different ;P
<pie__>
oh shi :D
<clever>
court order to dropbox, who has the file with hash X!
contrapumpkin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<clever>
this is where it comes into play, is the DDT based on the hash of the plaintext, or the hash of the ciphertext?
<pie__>
clever, oh crap i just realized
<pie__>
clever, i was just htinking about the arrangement of duplicate blocks
<pie__>
but you were talking about somethng completely different
<pie__>
the actual data
<clever>
yeah
<pie__>
if its of the ciphertext that should be fine i guess
<pie__>
if its of the plaintext, thats kinda f*****, without a salt. but i expect osmeone would have complained loudly
<pie__>
so, theres a good diagram...but still no clue what actually goes in the dedup table
<pie__>
i thiiink its the crypted data
<pie__>
because it talks about needing deterministic crypt results
<clever>
that sounds likely
<clever>
also, i think the DDT operates at a low level, where you just have a raw block storage
<clever>
you give it a block, and it returns a block-id
<clever>
if DDT is off, thats the location
<clever>
if DDT is on, it does a lookup (based on hash), and may return an old location
clacke_movim has joined #nixos
<clever>
filesystems and encryption are then built ontop of that, and dont decide where to write things, just "save this block", "where did you put it?"
<pie__>
sure
zeta_0 has quit [Quit: taking break]
<hyper_ch>
pie__: there's a talk from Tom Caputi where he first published the encryption. it has pretty good explanation.
<pie__>
clever, ...can i put a zfs dev on on a zfs block device and just fully encrypt the outer block device
<clever>
pie__: no idea
<clever>
definitely dont put a pools vdev onto a volume within the same pool :P
schjetne has joined #nixos
toppler`` has quit [Ping timeout: 245 seconds]
<ashkitten>
since yall are talking about zfs, have any of you had issues with nix not being able to allocate memory after a long uptime? (i assume due to zfs)
<ashkitten>
sometimes on my laptop i have to reboot before nix will work again..
<hyper_ch>
ashkitten: same here with notebook but that's probably due to free diskspace... I remember when free diskspace is < 20% it goes into kind of "panic mode" and I'm constantly below 20%
toppler`` has joined #nixos
<hyper_ch>
however, on 5 servers I have no such issues
<ashkitten>
hm
<hyper_ch>
the notebook just gets slow over time and when I reboot it then (after a week or so), it will first run a lot of txg_commit
<hyper_ch>
but to my knowledge the "panic mode" is still in existance
<ashkitten>
it's not that issue
<hyper_ch>
well, was just a guess
wfranzini has joined #nixos
wfranzini has quit [Remote host closed the connection]
Rusty1 has quit [Quit: Konversation terminated!]
mexisme_ has quit [Ping timeout: 268 seconds]
tobiasBora has joined #nixos
appleclusters has quit [Quit: Connection closed for inactivity]
sb0 has joined #nixos
o1lo01ol1o has joined #nixos
rauno has quit [Ping timeout: 252 seconds]
wfranzini has joined #nixos
wfranzini has quit [Remote host closed the connection]
o1lo01ol1o has quit [Ping timeout: 258 seconds]
hoijui has joined #nixos
orivej has joined #nixos
schjetne has quit [Ping timeout: 252 seconds]
palo1 has joined #nixos
palo1 is now known as palo
<clever>
pie__: the master key part is very much like luks
<clever>
pie__: and reveals something mildly scarry
<pie__>
scarry or scary
<pie__>
are you reading what i linked?
<clever>
pie__: since send|recv can cpy encrypted datasets, an attacker with physical access can clone your dataset, and use incremental sends to update his clones if he regains access
<clever>
reading the heckel.xyz blog
<pie__>
so yes
<pie__>
i havent actually started reading the text yet
<clever>
but, the attacker would only have the cihpertext, initiallly
<pie__>
clever, well..it's data _at rest_ for a reason i guess :/
<clever>
if he also copied the (encrypted) master key block each time, and keeps an eye on leaked passwords
<pie__>
hm
<clever>
and if your passphrase is ever leaked in the future, and the attacker has a copy of the master-key, encrypted with that, he can then break it
<clever>
even if you change the password immediately
<clever>
they have the old master block, and can effectively undo your password change
<ashkitten>
how is that different from sending *any* encrypted data
<clever>
ashkitten: thats the difference between changing the main key, vs wrapping the main key with a second passphrase
<ashkitten>
ah
<clever>
if you change the main key, then any future data is safe, but you have to re-encrypt every single block
<clever>
if you just protect the real key with a second passphrase, and your just re-encrypting it with a new passphrase
ghasshee has quit [Ping timeout: 248 seconds]
<clever>
then an attacker can undo a password change
<ashkitten>
well, i guess they should stress that security caveat
<clever>
luks has the same issue
<ashkitten>
but if you are compromised you should reencrypt it
<clever>
ashkitten: the problem, is what if an evil maid cloned your drive 3 years ago
schjetne has joined #nixos
<clever>
ashkitten: and then 2 years more down the road (5 years after the clone), your password was leaked, and you immediately changed the phrase in zfs
<pie__>
wait *stil thinking*
<clever>
an attacker with that initial clone, can just undo the password change next time they gain access
<clever>
and read any future data youve written
<clever>
you may not even know such a clone exists
<clever>
and think that changing the passphrase is enough
<pie__>
can you say again how they undo the change?
<ashkitten>
i will say you shouldn't be just changing your password if you're compromised, but i understand the premise
<clever>
pie__: the real key used for encryption of the entire disk, is encrypted with the passphrase, and stored somewhere on-disk
<pie__>
aha
<clever>
pie__: when you change your passphrase, your basically just doing encrypt(decrypt(cipher,oldpw),newpw)
<pie__>
and how would the correct way to do this be?
<ashkitten>
pie__: they "undo" the change by decrypting their dumped master key with your old passphrase, and using it to decrypt the current dataset
<pie__>
i dont see how its not master keys all the way down for the correct way
<clever>
yep
<clever>
the only solution is to change the master key at regular intervals
<ashkitten>
the correct way would be to create a new master key and reencrypt the entire dataset
<pie__>
right
<clever>
but that involves re-writting all data to disk
<pie__>
well
<pie__>
you gotta do what you gotta do
<clever>
for zfs, you could just make a new dataset, which doesnt inherit the key from another dataset
<clever>
and then maybe send|recv between the 2
<clever>
but that doubles your usage
<ashkitten>
i will say, for low-risk scenarios you probably will never run into this threat profile
<pie__>
clever, open a zol issue?
<clever>
seems like such an obvious problem, and no real cheap way to solve it, same reason luks has this issue
<clever>
luks also takes advantage of that, to add more features
schjetne has quit [Ping timeout: 250 seconds]
<clever>
rather then just storing a single encrypt(master,pw), it stores a list of them
<pie__>
meh, my scenario recently - which i havent actoually thought about yet - is, what if i was a journalist
<pie__>
id need to protect my sources n junk
<clever>
and any of those passwords can unlock the master key
<pie__>
but i dunno
<pie__>
well yeah, for the duration of the transfer, but nothing else makes any sense<clever> but that doubles your usage
<ashkitten>
if you have a threat profile that includes this, you probably have other issues. however, this is still something that should be addressed (at least an easy way to change the master key of a dataset)
isHavvyGhosting has joined #nixos
isHavvy has quit [Read error: Connection reset by peer]
<pie__>
and well, i guess for zfs you cant even say its just random bytes, so i guess the only real solution is to just wrap it in luks (i think you can have the header on a separate device or something, technically?)
<pie__>
i get into the crypto yak shave real easily as you might be able to tell
<clever>
yeah, luks has a detached header option
<pie__>
why leave broken what can be fixed? :P
<clever>
but also, with a detached header, the luks partition wont even have a uuid
<clever>
so there is no way to reasonably locate it, or even tell if its there
<pie__>
thats the point i guess
<clever>
so you may try to decrypt the wrong disk with that header, or accidentally format over it
ym555 has quit [Ping timeout: 248 seconds]
clacke_movim has left #nixos [#nixos]
clacke_movim has joined #nixos
<pie__>
clever, slightly tired rant:
<pie__>
this is the type of stuff that should be modeled in a crypto modeling language and just have it puke out all the possible weaknesses and attack scenarios m i rite?
jasonrm has joined #nixos
<hyper_ch>
clever: I wonder if that wouldn't be enough for protecting the master key: create an encrypt dataset: pool/encryption --> then create child sets of it that inherit they key properties pool/encryption/nixos -> however the nixos DS would not contain the master encryption key as it points out that this key in in pool/encryption --> you can then rawsend the nixos dataset but the master key won't be sent with it
Kiren has quit [Quit: Page closed]
<pie__>
clever, "Keys and key sources can be changed after the dataset/volume creation, and without re-encrypting the data (as they are never used directly). " so im actually reading the text now, what does that mean
<hyper_ch>
and you just have to make sure that you backup the pool/encryption dataset to secure devices (e.g. 2-3 usb thumb drive that are stored in your bank's vault
<clever>
pie__: the master key wont change, and it may be possible to undo a pw change
<pie__>
right but it says its never used directly
<clever>
never revealed to the user
<pie__>
oh
<clever>
so you cant just run a zfs command to expose it to the world
<pie__>
that makes more sense, thats the phrase i didnt really get
<pie__>
clever, the man page actually lists some possible weaknesses
<pie__>
" Since compression is applied before encryp‐
<pie__>
mation about which blocks are equivalent in a dataset and will incur an extra CPU cost per block written.
<pie__>
tion datasets may be vulnerable to a CRIME-like attack if applications accessing the data allow for it. Deduplication with encryption will leak infor‐
<pie__>
"
clacke_movim has joined #nixos
<pie__>
clever, i do recommend going through the diffed man page with something like meld <(man /nix/store/msrcfzv0i9k1dgj9dbpqkm9gnks1q7xy-zfs-user-0.8.0/share/man/man8/zfs.8.gz | col -b) <(man /nix/store/pf10i
<pie__>
5lvzqgg3b1jy00giz84zv05lplf-zfs-user-0.7.13/share/man/man8/zfs.8.gz | col -b)
<pie__>
the screen width is kind of annoying though i should fix that
<{^_^}>
[nixpkgs] @Mic92 opened pull request #62648 → aerc: init at 0.1.0 → https://git.io/fjuIp
<pie__>
looks like you probably want to actively use -w with encrypted sends
<pie__>
Note that if you do not use this flag for sending encrypted datasets, data will be sent unencrypted and may be re-encrypted with a different
<pie__>
#
<pie__>
encryption key on the receiving system, which will disable the ability to do a raw send to that system for incrementals.
<pie__>
Note that if you do not use this flag for sending encrypted datasets, data will be sent unencrypted and may be re-encrypted with a different
<pie__>
encryption key on the receiving system, which will disable the ability to do a raw send to that system for incrementals.
<pie__>
oops
gratto has joined #nixos
rauno has joined #nixos
xantoz has quit [Remote host closed the connection]
nschoe has quit [Read error: Connection reset by peer]
mexisme_ has joined #nixos
weedloser_ has quit [Ping timeout: 258 seconds]
mexisme_ has quit [Ping timeout: 268 seconds]
rprije has quit [Remote host closed the connection]
rprije has joined #nixos
roosemberth is now known as Orbstheorem
Tucky has joined #nixos
civodul has joined #nixos
<GlennS>
How do NixOS and NixOps handle kernel upgrades? Do they require a reboot? Or are they able to do it in place?
alp has joined #nixos
iqubic` has joined #nixos
orivej has quit [Ping timeout: 248 seconds]
iqubic has quit [Ping timeout: 252 seconds]
<LnL>
you need to reboot
davidak has joined #nixos
Ariakenom has joined #nixos
bahamas has joined #nixos
hmpffff has joined #nixos
<GlennS>
How do I know when I should be rebooting my servers?
<ar>
once they hit 60 days of uptime, they're ripe for a healthy reboot.
<GlennS>
OK, that's useful to know. I haven't been rebooting my servers. I think I should put up a second instance and a load balancer before I do.
lsyoyom has joined #nixos
<ar>
GlennS: i'm actually somewhat serious here; keeping things running for months or years makes things prone for vulnerabilities and configuration errors
<GlennS>
I'm just talking to my colleague here who has a server with over a decade of uptime :o
<andi->
oh boy.. Please tell me it is a pacemaker and not some sort of simple stupid web/mailserver
sestrella has joined #nixos
<GlennS>
Anyway, thanks for the advice, hopefully you have saved me from finding out the hard way.
ericsagnes has quit [Ping timeout: 257 seconds]
johanot has joined #nixos
nschoe has joined #nixos
ricekrispie has joined #nixos
Shoubit has joined #nixos
tjay has joined #nixos
Glider_IRC_ has quit [Ping timeout: 248 seconds]
goibhniu1 has joined #nixos
vidbina has joined #nixos
<tjay>
hi all i'm kinda new to NixOS and i was wondering how i can enter the configuration.nix file after the install
thc202 has joined #nixos
NinjaTrappeur has joined #nixos
<lucus16>
Use a text editor to edit /etc/nixos/configuration.nix
ericsagnes has joined #nixos
<marek>
nix-env --list-generation is empty for me, any idea how could that happen?
<{^_^}>
[nixpkgs] @peti pushed 6 commits to haskell-updates: https://git.io/fjut5
<johanot>
marek: if you want system generations, try: nix-env --list-generations --profile /nix/var/nix/profiles/system
<{^_^}>
[nixpkgs] @peti pushed 3 commits to haskell-updates: https://git.io/fjutd
<marek>
johanot: thank you! hmm, I was almost sure it did work without -p before
<tjay>
@lucus16, thanks for the reply
vidbina has quit [Ping timeout: 258 seconds]
<{^_^}>
[nixpkgs] @peti pushed 2 commits to haskell-updates: https://git.io/fjutA
ricekrispie has quit [Read error: Connection reset by peer]
mkoenig has quit [Ping timeout: 252 seconds]
ricekrispie has joined #nixos
sestrella has quit [Read error: Connection reset by peer]
eskytthe has joined #nixos
mkoenig has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed commit from @Infinisil to master « haskellPackages.heap: Fix build »: https://git.io/fjuqT
<{^_^}>
[nixpkgs] @volth opened pull request #62651 → websocat: add support for wss:// → https://git.io/fjuqk
<dminuoso>
gchristensen: Any particular reason why you say that old version of libjasper has security vulnerabilities? Do you have any resource at hand for me to look at?
<{^_^}>
[nixpkgs] @peti pushed 7 commits to haskell-updates: https://git.io/fjuqm
m0rphism has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed commit from @Infinisil to haskell-updates « haskellPackages: clean up some overrides »: https://git.io/fjuqs
<kraem>
how do i make sure to clean up after `nix-build <nixos/nixpkgs> -A .......` if the command failed and i cant delete the result symlink?
arjen-jonathan has joined #nixos
<kraem>
(i can't delete it because it was never created)
<yorick>
adisbladis: wait, buildGoModule also exists, what's the advantage of vgo2nix?
<nschoe>
kraem, if the result symlink was not created, then there's no root for what build, downloaded and put in the store, so a nix-collect-garbage should clean it.
ericsagnes has joined #nixos
<kraem>
nschoe: ok, thanks!
<adisbladis>
yorick: I guess nothing at this point
shad has quit [Remote host closed the connection]
<nschoe>
kraem, don't use the -d option, though.
shiver has joined #nixos
shad has joined #nixos
<kraem>
as it won't remove anything built with the command or as it removes old generations as well?
<nschoe>
kraem, the latter.
<kraem>
nschoe: ok thanks :)
jaeckel has joined #nixos
gratto has joined #nixos
linarcx has joined #nixos
ericsagnes has quit [Ping timeout: 272 seconds]
<Miyu-saki>
Any thoughts on sharing my /nix/store with a transient qemu guest?
<Miyu-saki>
Rather, any suggestions
<Miyu-saki>
I'm thinking of setting up a temporary user.
WizBright has quit [Ping timeout: 244 seconds]
WizBright has joined #nixos
sgo_ is now known as sgo
j4m3s1 is now known as j4m3s__
<Miyu-saki>
Hm, OTOH, I can just generate a privkey and use the same privkey for the same guest.
jaeckel has quit [Remote host closed the connection]
<makefu>
anyone using logstash? i tried to set logstash.plugins = [ pkgs.logstash-contrib ]; as described in the options, however i get "NameError: `@path.plugins' is not allowable as an instance variable name"
<{^_^}>
dminuoso: If a Nix file foo.nix starts with something like `{ stdenv, cmake }:`, you can build it with `nix-build -E '(import <nixpkgs> {}).callPackage ./foo.nix {}'`
<dminuoso>
infinisil: Ah cheers!
<arianvp>
Can someone explain to me in simple terms why we need /run/opengl-driver ?
knupfer has joined #nixos
psy3497 has quit [Ping timeout: 246 seconds]
ng0 has joined #nixos
<tilpner>
arianvp: AFAICT so that there's a fixed location for the libraries. Having different locations per driver would mean needing to build graphical packages once per driver for barely any benefit
vidbina has quit [Ping timeout: 248 seconds]
jackdk_ has joined #nixos
hoijui has quit [Ping timeout: 252 seconds]
xvnvx has joined #nixos
fendor has joined #nixos
xvnvx has quit [Excess Flood]
linarcx has quit [Quit: WeeChat 2.4]
<Miyu-saki>
... I just decreased the testing time from ~10 minutes to ~2 minutes by using a local binary cache. Why didn't I think about this before. -_-
<emilsp>
hmm, the npm package that one gets with nodejs_11_x is fairly old - is there a nixy way I can get a newer npm version?
xvnvx has joined #nixos
knupfer has quit [Ping timeout: 248 seconds]
sigmundv has joined #nixos
sveitser2 has joined #nixos
ericsagnes has joined #nixos
[Leary] has joined #nixos
<{^_^}>
[nixpkgs] @Profpatsch opened pull request #62656 → purescript: add test that compiles & runs a minimal module → https://git.io/fjuOV
<arianvp>
tilpner: but isn't the whole point of libglvnd and vulkan-loader that you can link against those and it is their task to dispatch to the right library?
<arianvp>
I would understand if only those would be in /run/opengl-driver but there's a lot more there
<arianvp>
Seems to make everything more impure than needed
ottidmes has joined #nixos
Lears has quit [Read error: No route to host]
Cale has joined #nixos
<ottidmes>
I have this piece of nodejs software that I can start/restart/stop by passing that keyword to the file, like: node file.js start (it starts a service in the background and exits). Instead of doing this manually I was thinking to have this be managed by a systemd service, what is the best way to achieve this?
<dminuoso>
Mmm. So I have a derivation that depends on a library in a different version than supplied by nixpkgs. What's the recommended mechanism to deal with this?
<ottidmes>
Guess I could just configure all cases like ExecStart and ExecStop and set the type to forking, but seems like a common enough thing to have a nicer solution for
<dminuoso>
(The derivation is under my control, so..)
<ottidmes>
dminuoso: override the lib at callPackage with the specific version you need
johanot has quit [Ping timeout: 248 seconds]
<dminuoso>
ottidmes: Ah so I'd include a lib-in-my-version.nix in my repo and just callPackage it
<Miyu-saki>
Wait, what, nixos-rebuild does not honor NIXPKGS_CONFIG?
<ottidmes>
dminuoso: you would do something like my-package = callPackage ./path/to/package.nix { lib = callPackage ./path/to/lib/with/custom/version.nix { }; };
<Miyu-saki>
Hm, so it works if you used `sudo su`
<dminuoso>
ottidmes: Ahh okay. I need to stop doing these `{ pkgs ? import <nixpkgs> {} }: ...` things and start parametrizing my derivations over packages. :)
eskytthe has quit [Ping timeout: 258 seconds]
<Miyu-saki>
But `nixos-rebuild switch` doesn't work either way.
<Miyu-saki>
:(
<Miyu-saki>
So a few things here, I guess. `sudo` doesn't preserve NIXPKGS_CONFIG and `nixos-rebuild` doesn't either, for some reason.
buffet_ is now known as buffet
<ottidmes>
Miyu-saki: don't you need sudo -E for that?
<Miyu-saki>
ottidmes:I feel like NIXPKGS_CONFIG should be preserved by default.
<Miyu-saki>
I mean, it does that for NIX_PATH for example.
<Miyu-saki>
Okay, apparently, I was wrong. So why does NIX_PATH get set but not NIXPKGS_CONFIG?
ng0 has quit [Quit: Alexa, when is the end of world?]
<Miyu-saki>
One thing that low-key annoys me is that I don't know what part is handled by the nix-daemon and what part is handled by the executable TBH.
<Miyu-saki>
`nix copy` for example.
<Miyu-saki>
(This is tangential to the topic, just wanted to say that lol)
<Miyu-saki>
Hm. Back to the problem at hand though. Why does `nixos-rebuild` not honor NIXPKGS_CONFIG?
bahamas has joined #nixos
<Miyu-saki>
Giff me a few.
<Miyu-saki>
[myrl@myrl:/etc/nixos/nixpkgs/nixos]$ nix-build '<nixpkgs/nixos>' -A system
<Miyu-saki>
trace: warning: Multiple directories is currently beta
<Miyu-saki>
error: undefined variable 'dmenu_sudo' at /etc/nixos/configuration.nix:263:45
<Miyu-saki>
So I guess it's in the mechanism of <nixpkgs/nixos> and not the nixos-rebuild tool.
rprije has quit [Read error: Connection reset by peer]
<Miyu-saki>
So that's why NIXPKGS_CONFIG is not honored.
<{^_^}>
[nix] @edolstra pushed to flakes « shell.nix: Use clang by default »: https://git.io/fju3X
<tobiasBora>
Hello,
<tobiasBora>
I'm trying to play with nix, and I can't find how to load a library in a shell (a binary file I have need the libstdc++.so.6 library). I tried to do nix-shell -p libstdcxx5 and nix-shell -p gcc, but none of them works
<Miyu-saki>
tobiasBora: Different search paths. You'll have to emulate an FHS for that.
<Miyu-saki>
I guess change /etc/nix/nixpkgs-config.nix itsefl?
<glenndavy>
hey there all, i’m needing shell.nix to be able to provide ‘sh’ to users in that shell. I need to include bash-interactive-5.0-p7 in the build inputs so that the shell installs it, but that ins’t a valid thing to go in build inputs array.. any ideas how i solve this?
civodul has joined #nixos
<exarkun>
_and_ my xmodmap configuration got dropped?
<{^_^}>
[nixpkgs] @mmahut opened pull request #62657 → kronometer: adding homepage and description → https://git.io/fju39
<exarkun>
how does that even
<Miyu-saki>
tobiasBora: Also, we're talking about an executable compiled for non-Nix, right?
drakonis has joined #nixos
<tobiasBora>
Miyu-saki: hum... yes, I'm talking about non-nix executables. I'm trying to understand the different methods to run them, and I'm interested by the "manual" method as well to understand what's happening. I tried to use LD_LIBRARY_PATH to force the library path, but it also fails:
<tobiasBora>
=> error while loading shared libraries
<{^_^}>
[nixpkgs] @Profpatsch merged pull request #62656 → purescript: add test that compiles & runs a minimal module → https://git.io/fjuOV
<{^_^}>
[nixpkgs] @Profpatsch pushed to master « purescript: add test that compiles & runs a minimal module »: https://git.io/fju3b
<gchristensen>
oh cool PR, Profpatsch ^
domogled has joined #nixos
drakonis_ has quit [Ping timeout: 258 seconds]
<tobiasBora>
hum, actually LD_LIBRARY_PATH=... ldd thescript seems to fix the library libstdc++. There is one line alone linux-vdso.so.1 (0x00007ffef3dfa000)
<tobiasBora>
maybe I also need to provide it
<joepie91[m]>
from the Nix documentation --- builtins.seq e1 e2 : Evaluate e1, then evaluate and return e2. This ensures that a computation is strict in the value of e1.
<joepie91[m]>
I have no idea what this is supposed to mean
<simpson>
joepie91[m]: You can comfortably discard the second half if you don't normally worry about Nix's lazy evaluation model.
<tobiasBora>
Miyu-saki: Hum... Does the ld-linux... forwards the environment?
<{^_^}>
Couldn't find in any packages
zupo has joined #nixos
<gchristensen>
simpson: why would somebody worry about the model?
werner291 has joined #nixos
<simpson>
gchristensen: For example, the fancy knot-tying recursion that we use for package scopes relies on Nix's lazy evaluation.
<joepie91[m]>
simpson: but I do. I want to understand the "why", it's just not written in a way that makes any sense to me whatsoever.
hoijui has joined #nixos
eskytthe has quit [Ping timeout: 246 seconds]
<simpson>
joepie91[m]: It might make too much sense to me. Do you feel comfortable with lazy vs. strict?
<gchristensen>
simpson: I guess I'm (and joepie91[m] is) looking for something more concrete
<joepie91[m]>
simpson: I know what 'lazy' means, and I understand the general lazy evaluation model in Nix, but 'strict' does not register to me as something that is an opposite of that, no
<simpson>
gchristensen: seq is merely a useful primitive that comes up often in lazy-by-default languages. There's not really any other way to strictify an expression, or to force an evaluation order without a data dependency.
<joepie91[m]>
is it just a synonym for 'eager'/'forced' evaluation?
JosW has joined #nixos
<tobiasBora>
Miyu-saki: hum, when I patch with patchelf, it actually works O_o What dark magic happens?
<joepie91[m]>
okay, so it's about forcing evaluation order by forcibly evaluating something that would otherwise have been evaluated lazily?
vidbina has joined #nixos
<dminuoso>
joepie91[m]: It's essentially about evaluation strategies.
<dminuoso>
joepie91[m]: call-by-name as opposed to call-by-value
<simpson>
joepie91[m]: Yes. However, note that the seq itself is lazy, once created. seq only promises that, once it starts being evaluated, it will evaluate all of e1 before starting on its actual value e2.
acarrico has joined #nixos
<joepie91[m]>
simpson: "all of" is only true for deepSeq, right?
<joepie91[m]>
with seq only forcibly-evaluating shallowly
<dminuoso>
joepie91[m]: In the Haskell world we are very careful to not conflate non-strict with lazy. They are different ideas.
<dminuoso>
So non-strict does not mean lazy.
<dminuoso>
(Lazyness is a possible implementatoin of non-strictness)
<tobiasBora>
Miyu-saki: ok, problem solved. It was a stupid error, I was not using ./ when i was removing the interpreter ^^' sorry, now it works!
<simpson>
joepie91[m]: I was hoping to paper over "normal form" with my phrasing. I'm not sure exactly how it works in Nix.
<joepie91[m]>
dminuoso: great, but that doesn't really help for my understanding here :)
<joepie91[m]>
I'm more interested in an explanation that can be intuitively understood, than one that is 100% compliant with a particular shared jargon
<tobiasBora>
Miyu-saki: You where talking about FHS, do you know if it's possible to run a FHS shell without actually writing a shell.nix file?
j4m3s__ has quit [Read error: Connection reset by peer]
<dminuoso>
joepie91[m]: strictness/non-strictness is easily explained in terms of graph reduction.
<joepie91[m]>
simpson: hmm, right, thanks
<Baughn>
Can't explain without explaining normal forms, I think.
<Baughn>
Which begs the question of what a 'normal form' is, for Nix.
<joepie91[m]>
dminuoso: 'graph reduction' is meaningless to people without an academic background, generally
<joepie91[m]>
so that is really not 'easily' at all
<simpson>
It's definitely not a concept that I could imagine making intuitive; nothing's intuitive about computer science.
* gchristensen
wishes for an explanation motivation
<dminuoso>
joepie91[m]: You do not need an academic background.
<gchristensen>
I get what seq _does_ I just don't really get _why_ it is wanted.
IuliusCaesar has joined #nixos
<andi->
is that another trivial thing? :)
<Baughn>
gchristensen: Because it's fairly easy to get space leaks without it.
<{^_^}>
[nix] @JorisE opened pull request #2918 → Minor typo → https://git.io/fjusE
<Baughn>
The canonical example would be foldl vs. foldl'.
<simpson>
gchristensen: Because there's not really any other way to get its behavior; it's primitive, and somebody found that they wanted it at some point.
<joepie91[m]>
simpson: can you name a concrete example of a case where a sequential evaluation guarantee is necessary, but there is *not* a data dependency that makes it so?
<dminuoso>
joepie91[m]: So if you think of an expression as a graph of application nodes, then something like `foo (bar quux)` has two ways of reduction.
<dminuoso>
joepie91[m]: You either reduce `bar quux`, call the resulting value `q`, and then you reduce `foo q`, or you start with the `foo (...)` part.
<gchristensen>
simpson: that does not explain why it is wanted
<Baughn>
joepie91[m]: Summing a large list with fold. If you don't use seq you'll get an equally large chain of thunks, which may stack-overflow when evaluated.
<simpson>
gchristensen: I mean, I'm not going to balk if there's an RFC to remove it. I'm just trying to explain that it's here *now* almost certainly because of a desire in the *past*, and that desire was likely tied to the fact that seq can't be hand-crafted otherwise.
<joepie91[m]>
Baughn: how does seq solve this, and why can this not be addressed on a runtime level?
<gchristensen>
simpson: this is not progressing :P
<dminuoso>
joepie91[m]: the runtime is not an all knowing oracle.
<joepie91[m]>
(because that feels like a workaround :P)
srl295 has quit [Quit: Connection closed for inactivity]
<simpson>
joepie91[m]: By definition, that necessity would be a data dependency, right? seq isn't needed for computational completeness AFAIK, just for controlling evaluation order.
<Baughn>
joepie91[m]: seq lets you say "Before returning (a + b + (...)), first evaluate (a + b).
<marek>
any idea how to make profiles to source /etc/profile?
<joepie91[m]>
simpson: "gchristensen: seq is merely a useful primitive that comes up often in lazy-by-default languages. There's not really any other way to strictify an expression, or to force an evaluation order without a data dependency."
<joepie91[m]>
that is what I'm deriving my question from
rprije has quit [Read error: Connection reset by peer]
<simpson>
joepie91[m]: Baughn and dminuoso have been trying to explain one such case, where a large chain of additions could explode in heap space usage without it.
<gchristensen>
I thought FP was about freeing yourself from the physical computer or whatever
<Baughn>
Given the list you're adding can be generated lazily, this could take an O(1) space algorithm to O(n).
<gchristensen>
(that was more reductionist than I intended :x)
<Baughn>
gchristensen: Oh, sure. This makes no semantic difference.
<dminuoso>
gchristensen: In fact `seq` is a dirty little primitive that breaks so many nice properties of Haskell.
<dminuoso>
gchristensen: Which is why we even go to great lengths pretending it doesn't exist when applying formal methods.
<Baughn>
dminuoso: Let's not tell him about unsafePerformIO.
<gchristensen>
hah!
<gchristensen>
I know about unsafePerformIO :P
<gchristensen>
(I even know a little Haskell, despite pretending not to)
<simpson>
gchristensen: We could switch to bounded computation. Let's write everything as a stream processor taking at most quadratic time~
<gchristensen>
yes!
<gchristensen>
perfect
<dminuoso>
gchristensen: The Haskell Report takes great care not to talk about evaluation order in the slightest.
<simpson>
(Hope you don't need to multiply matrices!)
<dminuoso>
So even the `seq` primitive does not give you the same guarantees as in nix.
<joepie91[m]>
Baughn: thanks, that seems like a reasonably accessible explanation. reading it now
<dminuoso>
That is, in Haskell `p seq q` does not guarantee p will be evaluated before q.
<gchristensen>
lol
<dminuoso>
Because that would require talking about evaluation order.
<joepie91[m]>
as an aside: there are plenty of people who have never touched Haskell, so it's probably not a great idea to assume Haskell knowledge when explaining concepts of Nix to people :)
<Baughn>
I persist in thinking the Nix language was a mistake.
<gchristensen>
spicy
<Baughn>
It's very similar to google's BCL.
<Baughn>
And that's not a compliment. XD
<Baughn>
The best that can be said (about both of them) is they _do the job_.
<simpson>
Nix is so much nicer than BCL.
<lucus16>
Baughn: in what way is it a mistake and how would you have solved that?
<dminuoso>
Going back to fixing stupid packages. Sighs, nix can be annoying. :(
* joepie91[m]
is actually fairly happy with Nix the language, aside from some papercuts
<simpson>
It's like complaining that a vulture is bald; they have a good reason for it!
<Baughn>
lucus16: First of all, it lacks static typing -- and that's cost me so very much time. Even adding static types to Nix would be a massive improvement.
<joepie91[m]>
please no :)
<Baughn>
I'm just not sure that's doable without a redesign. You'd have to use pretty exotic types.
<Baughn>
(Row types, structural subtyping... I dunno, I'm not a type theorist.)
acarrico has quit [Ping timeout: 268 seconds]
<dminuoso>
Baughn: You can just use Dhall..
<dminuoso>
I mean some typing is better than no typing.
<lucus16>
Baughn: The typescript type system might be a decent match for Nix
<Baughn>
dminuoso: Didn't know about it. Hmm.
<Baughn>
I've dreamt of writing something like that.
<Baughn>
lucus16: Also my thought, but not a *perfect* match. A custom type system would be best, except...
<dminuoso>
Baughn: There's a dhall2nix tool to convert your dhall expressions into nix.
<Baughn>
--again, not something I think I could do in a reasonable time.
<Baughn>
Anyway, since I'm ranting about Nix, next problem: Performance.
<Baughn>
It might be partially due to the lack of typing, but mostly it's the lack of compilation. `nixos-rebuild` takes 30-40 seconds before it even starts downloading packages.
<Baughn>
A solution there would have to look like SBCL, though. A partially compiled language that won't bother to compile the files _you_ provide it.
<joepie91[m]>
pretty sure there's a lot of low-hanging optimization fruit left in Nix
<niksnut>
I think the main issue about the language is that it has functions. It's a mistake to have a DSL with functions.
<niksnut>
Because any abstractions you might want to make with functions would be better expressed as DSL constructs.
<niksnut>
Also the performance problems stem from having to evaluate a whole purely functional program.
eskytthe has joined #nixos
<dminuoso>
niksnut: I doubt that has anything to do with it.
weedloser has joined #nixos
<Synthetica>
Are there already tests in place for Wayland compositors, only for Xorg?
<Synthetica>
s/only /or only/
<gchristensen>
dminuoso: oh?
<{^_^}>
[nixpkgs] @joachifm pushed commit from SLNOS to release-19.03 « firefoxPackages.tor-browser: 8.0.9 -> 8.5.0 »: https://git.io/fjuGU
hoijui has quit [Quit: Leaving]
Ariakenom has quit [Quit: Leaving]
<joepie91[m]>
Baughn: right, so that article made sense for the most part. to paraphrase in entirely different and intentionally simplified/analogized terminology, it will sort of "prime the cache" by evaluating a Thing before (partially-evaluating) the expression in which the Thing is used, so that the expression will contain the actual value of the Thing, instead of a pile of code that would compute that value later... so that the
<joepie91[m]>
internal representation of the expression doesn't become too big.
<joepie91[m]>
correct?
<joepie91[m]>
(conceptually, that is)
<simpson>
If that makes sense to you, then sure. There's no cache or partial evaluation AFAIK, but I don't know anything about this part of Nix internals.
eskytthe has quit [Ping timeout: 248 seconds]
<dminuoso>
joepie91[m]: Think of it as `foo (expensive computation)`. If `foo` is strict in its argument, it has to evaluate the `expensive computation` before it can enter `foo`
<dminuoso>
joepie91[m]: Now keeping this not-evaluated computation around requires some overhead. If we know with absolute certainty that this argument will be used, then its cheaper to just force the evaluation rather than keep some data structure around to represent this unevaluated expression.
<{^_^}>
[nixpkgs] @joachifm opened pull request #62658 → eudev: use utillinuxMinimal → https://git.io/fjuG6
<exarkun>
I put `VIRTUALENV = /tmp/some/path;` and `shellHook = "export PATH=$VIRTUALENV/bin:$PATH"` into my default.nix (which maybe I am abusing and I should use shell.nix? dunno)
<exarkun>
And I got a PATH with a nix store path in it! What's going on?
<{^_^}>
[nixpkgs] @jarlg opened pull request #62659 → nixos/undervolt: correct description of numerical input → https://git.io/fjuGP
<{^_^}>
[nixpkgs] @joachifm opened pull request #62660 → lvm2: cleanup & fix eval error when systemd=null → https://git.io/fjuGX
drakonis_ has joined #nixos
<dminuoso>
This is brilliant. So the authors took all the shared objects from various libraries and all tossed them into the top level directory of this package.
MmeQuignon has quit [Read error: Connection reset by peer]
drakonis has joined #nixos
zupo has joined #nixos
zupo has quit [Client Quit]
halfbit has quit [Quit: WeeChat 2.4]
Neo-- has joined #nixos
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62664 → pythonPackages.flask-swagger-ui: init at 3.20.9 → https://git.io/fjucy
infinee has joined #nixos
<infinee>
Good evening, anyone know why when I remote ssh into my nixos, it's not setting up my nix profile? I can't even run 'ls' from the ssh session because it thinks nothing is installed.
werner291 has joined #nixos
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62665 → pythonPackages.pytest-flask: init at 0.15.0 → https://git.io/fjucQ
<lordcirth__>
infinee, did you enable sshd in configuration.nix with "services.sshd.enable = true;" or some other way?
<infinee>
"services.openssh.enable = true;"
<hyper_ch>
infinee: configuration.nix helps to figure things out :)
<Baughn>
infinee: Are you doing anything special from the client side? Does .ssh/config contain interesting lines?
<Baughn>
*Bypassing* the default login profile is possible, given the right flags.
<infinee>
Baughn: I don't have a ~/.config/ssh dir
<infinee>
should it be openssh or sshd I'm enabling?
<Baughn>
sshd is an alias for openssh.
<Baughn>
infinee: And it's just ~/.ssh
MmeQuignon has joined #nixos
<Baughn>
SSH is much older than the .config convention.
<infinee>
Baughn: I only have some public keys in ~/.ssh
Mateon2 has joined #nixos
<infinee>
dont' have ~/.ssh/config file either
Mateon1 has quit [Ping timeout: 245 seconds]
Mateon2 is now known as Mateon1
<Baughn>
Right. Well, hand us your configuration.nix please.
<Baughn>
infinee: Also, is your homedir migrated from a non-NixOS system by any chance? The one on the server, I mean.
<infinee>
It shouldn't be. It was nixos installed from scratch
<infinee>
all my old dotfiles I grabbed from git, but I'm using fish on this so my .bashrc shouldn't matter
<Baughn>
Oh, that might be it. Fish, I mean.
<Baughn>
Do you have 'programs.fish.enable = true' in your configuration.nix?
<Baughn>
Aka. "
<Baughn>
"Please pastebin it for us."
<Baughn>
Your shell _needs_ to run /etc/profile, or the equivalent, and if fish doesn't do that by default then there you go. If you have the option of either setting 'programs.foo.enable' or adding 'foo' to environment.systemPackages, always do the former; don't do the latter.
<Baughn>
It doesn't look like fish is bash-compatible, so it probably doesn't execute /etc/profile.
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62666 → python: azure-keyvault: init at 1.1.0 → https://git.io/fjuCv
<adisbladis>
Also: Fish on Nix requires some special massaging (taken care of by programs.fish.enable on NixOS)
<warbo>
hello, I've updated to 19.03 but my custom Emacs (lucid toolkit) is crashing
<warbo>
it looks like it might be a known problem, to do with passing "-z nocombreloc" to the linker
<infinee>
ahh thanks Baughn, that worked!!
<infinee>
no it's not a laptop actually. it's my old development rig
<Baughn>
With Optimus? Huh.
<warbo>
I'm not sure where to look for that; would it be in the build script, as a gcc arg, or something?
<tilpner>
adisbladis: Huh, I thought there was. Guess I misremembered that
<warbo>
(I've very rarely used C)
<infinee>
Baughn: not sure how that got in there. I'm using a gtx10709
<infinee>
-9
MmeQuignon has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62667 → python: sh: fix (patch a test to be able to install again) → https://git.io/fjuCc
vidbina has joined #nixos
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62668 → python: python-redis-lock: init at 3.3.1 → https://git.io/fjuCg
<{^_^}>
[nixpkgs] @fpletz pushed to staging « pulseaudio: fix alsa due to alsaLib headers move »: https://git.io/fjuC2
Jackneill has joined #nixos
<infinee>
where/how do I configure sshd? /etc/static/ssh/sshd_config seems to be readonly (I'm guessing for a reason)
shibboleth has quit [Quit: shibboleth]
<clever>
infinee: services.openssh in configuration.nix
linarcx has joined #nixos
<infinee>
ahh programs.ssh.forwardX11=true; was what I was looking for
<{^_^}>
[nixpkgs] @teto opened pull request #62669 → Add one latex and one lua LSP server → https://git.io/fjuCK
drakonis_ has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @vanschelven opened pull request #62670 → pythonPackages.flask-swagger: init at 0.2.14 → https://git.io/fjuCX
<tjay>
So i was wondering if anyone can tell me how Gaming on NixOS is, reason i'm a streamer and i really like NixOS because 1. it's a Dutch Distribution -- 2. I like the Nix approach
drakonis_ has joined #nixos
<clever>
tjay: obs and steam both work for me
<slabity>
tjay: Steam works fine. Proton too
<Baughn>
tjay: Steam works fine, most of the time. Some (C#-based, typically) games may need extra work.
<genesis>
assaultcube forever
<Baughn>
"NixOS is Dutch" is a poor reason to select an OS, though. ;P
<tjay>
Baughn, i have to support my fellow dutch guys tho :P
<ryantrinkle>
getting this: ! LaTeX Error: File `tikz.sty' not found.
<matthewbauer>
Izorkin: looks good! thanks for reminding me
<Izorkin>
matthewbauer: thanks!
<o1lo01ol1o>
I have a shellHook expression that sets 3 environment variables. when I enter the shell, one is correctly set, another is not set at all, and the third, LD_LIBRARY_PATH, is not set at all. What gives?
<ryantrinkle>
i'm not too familiar with latex; anyone know how I can get tikz.sty in the right place?
thomasd has quit [Quit: Page closed]
<jomik>
Thanks! I'll try that.
<tilpner>
ryantrinkle: If you can spare the bandwidth, try scheme-medium instead
<infinee>
thanks again Baughn and clever , I can ssh to my nixos and fwd'ing X11 now
<infinee>
can get rid of extra keyboard/mouse on my desk lol
averell has joined #nixos
<Izorkin>
to all - please check PR #44343 How to need to merge?
arjen-jonathan has quit [Ping timeout: 248 seconds]
clacke_movim has left #nixos [#nixos]
<{^_^}>
[nixpkgs] @kfiz opened pull request #62671 → tasks/networking-interfaces.nix: Enable ip_forwarding for ipv4 and p… → https://git.io/fju8I
clacke_movim has joined #nixos
mmlb6 is now known as mmlb
emptyflask has quit [Ping timeout: 248 seconds]
wfranzini has quit [Remote host closed the connection]
isHavvyGhosting has quit [Read error: Connection reset by peer]
<jomik>
tilpner, clever: when I use `callPackage = pkgs.newScope pkgs'`, it does not seem that the callPackage that is passed around also carries this new scope? Some of my expressions use callPackage, and they also need access to pkgs'.
<kraem>
Can't find anything in the manual on ZFS. What I'm trying to do is dual boot macOS and nixOS. NixOS being on ZFS + LUKS encryption. Do you think i could follow this (accustomed to my setup of course: https://bpaste.net/show/f22128759758)https://nixos.wiki/wiki/Full_Disk_Encryption#zimbatm.27s_laptop_recommendation - stopping after the cryptsetup commands and continuing here
<{^_^}>
[nixpkgs] @matthewbauer pushed 5 commits to master: https://git.io/fju8F
<Izorkin>
matthewbauer: thanks!
shibboleth has joined #nixos
Glider_IRC has quit [Ping timeout: 248 seconds]
<werner291>
Hmmm... So I have this git repo with a single shell script in it. I've been trying to get that script to run periodically, but I'm not quite sure where to start...
rprije has quit [Remote host closed the connection]
rprije has joined #nixos
arjen-jonathan has joined #nixos
<inquisitiv3>
I'm getting a build error when I try to build my config for Home Manager, and can't figure what I'm doing wrong. Could one of you please take a quick look and see what I've done wrong? The error message is included in the snippet description.
<{^_^}>
[nixpkgs] @matthewbauer pushed to master « systems: fixup from last commit »: https://git.io/fjuBt
ilmu has quit [Ping timeout: 258 seconds]
<acowley>
I am building an executable for ARM (Raspberry Pi) on an x86_64 machine. I now want to strip the executable on the build platform before copying it to the target machine. This used to work using the regular strip from binutils on the build machine, but it no longer does. Is there a way to build a strip executable that knows how to understand ARM executables?
jluttine has quit [Quit: WeeChat 2.4]
simon_weber has joined #nixos
jasongrossman has quit [Ping timeout: 245 seconds]
<acowley>
It cuts about 2MB from this executable, so is a minor nice thing to have working.
<matthewbauer>
acowley: yeah binutils is target specific
<matthewbauer>
you need `(import <nixpkgs> {crossSystem = {...};}).buildPackages.binutils`
<acowley>
matthewbauer: Ah, thank you! I had almost gotten there, but then got myself turned around.
<simon_weber>
I'm looking to deploy a custom python server application. Currently it's not set up as an installable python package -- the source is just copied to the server. Is there an equivalent approach in nix, or do I have to make it a real package and add versioning, etc?
<acowley>
strip is only in binutils-unwrapped for some reason, and it's buried under a platform directory, but it all works now.
<acowley>
The cross compilation chapter of the manual tripped me up because it made me want to specifically set buildPlatform, hostPlatform, and targetPlatform
<acowley>
But I couldn't do that
<acowley>
I already happened to have a crossSystem configured nixpkgs in my .nix file, but I missed the point that the buildPackages attribute would have hostPlatform set to my buildPlatform.
<marek>
adisbladis: hm, github is saying changes are requested, but I do not see how to make sure to confirm it is provided
toppler`` has quit [Ping timeout: 248 seconds]
Glider_IRC has quit [Quit: Leaving]
judson has joined #nixos
Glider_IRC has joined #nixos
<{^_^}>
[nixpkgs] @Ma27 merged pull request #62165 → trinity: small clean up → https://git.io/fjRxt
Rabautius has quit [Remote host closed the connection]
camsbury has quit [Remote host closed the connection]
camsbury- has quit [Remote host closed the connection]
<kraem>
clever: in the justdoit.nix script you sent me you're creating a swapdevice, would it be possible to have a swap file within one of the zfs filesystems or create a separate zfs filesystem for it?
shabius has joined #nixos
<gchristensen>
are you aware of the hazards of swap on zfs?
Soo_Slow has quit [Remote host closed the connection]
<bjpbakker>
Hi, since about a week I get a lot of 'HTTP error 200 (curl error: Failure when receiving data from the peer)' errors during a system rebuild. I am on the unstable channel. Is this some known issue?
<clever>
kraem: what gchristensen said ^^
<kraem>
gchristensen: i am not
orivej has quit [Ping timeout: 272 seconds]
<kraem>
but it's safe to create a swap "partition" or filesystem under the zfs pool as you did in your script? I just thought it would be nice to not determine the amount of swap i want/need while paritioning (i basically just want it to suspend+hibernate = because bad batter in laptop)
<clever>
kraem: my swap device on lvm on luks
andymandias has joined #nixos
<clever>
so zfs is being bypassed
<clever>
but with recent improvements to nixos, you can just make 2 luks devices, one for zfs, one for swap, and share the passphrase
srl295 has joined #nixos
weedloser has quit [Quit: Leaving]
<kraem>
ah that's why you're running LVM? was hoping i'd be able to skip that abstraction layer (to remember what i've done later :P)
<kraem>
would i be prompted to enter the password for both luks volume on startup in that case (creating 2 LUKS devices)?
<colemickens>
Anyone have a fontconfig in nix config that they're particularly proud or or happy with?
ambro718 has joined #nixos
<clever>
kraem: thats the recent improvement to nixos
<clever>
kraem: it will remember the passphrase, and try it on the 2nd luks device for you
<clever>
which is why i previously used lvm
bjpbakker has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #62606 → nixos/grub: Add defaultText for font option → https://git.io/fjE5E
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fjuRV
gml8 has joined #nixos
knupfer has joined #nixos
knupfer has quit [Client Quit]
gentauro has joined #nixos
<simon_weber>
I'm confused about how paths are handled: why is it that `./dir` and `./file.foo` are valid, but `./` and `.` give a syntax error? Strangely, `./dir/..` appears to work for referring to the current directory.
<samueldr>
though it's not exactly written that it must not end with a slash
Orbstheorem has left #nixos ["WeeChat 1.8"]
<simon_weber>
gotcha; thanks
shibboleth has quit [Quit: shibboleth]
<samueldr>
relatedly, that chapter (15 currently) "Nix Expression Language" has been extremely useful when I was learning how to use and understand Nix
<kraem>
oh btw; the ashift option, how do i find out what the sector size is on my ssd? or is it standard to use 12?
<clever>
kraem: fdisk -l /dev/sda
<slabity>
kraem: 12 bytes for a sector does not sound standard
Shoubit has quit [Ping timeout: 245 seconds]
<lordcirth__>
slabity, ashift=12 means 2^12
<lordcirth__>
It's a bitshift
fendor has quit [Ping timeout: 258 seconds]
<lordcirth__>
12=4k sectors
<IslandUsurper>
Alright, from the NixOS manual, I read, "Warning: Unfortunately, Nixpkgs currently lacks a way to query available configuration options." What are my other options? Inspecting the expression in the source repo?
<slabity>
IslandUsurper: For NixOS options? Use `man configuration.nix`
<IslandUsurper>
Sorry, it's for a package's options, slabity.
<kraem>
gchristensen: yeah i found it at open-zfs.org - guess i'll just trust that source then :) - but could anything go wrong or would i just lose performance?
<realloc>
btw. does anyone here use the nextcloud module? If so, could you share your config? I cant get the connection to the (postgresql) database to work
domogled has quit [Ping timeout: 258 seconds]
fendor has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @Infinisil merged pull request #58552 → nixos/deluge: add autFile, config and port options → https://git.io/fjTdu
<kraem>
tilpner, clever, gchristensen Thank you so much for your help. I'm up and running on my macbook with zfs+luks now :) Alot of tweaking left but the hard part is over - so nice to have a reproducable setup!
<kraem>
clever: can't remember exactly what the nixos-install command said but i ended up commenting out 'boot.suppertedFileSystems = [ "zfs" ];`
civodul has quit [Quit: ERC (IRC client for Emacs 26.2)]
clacke_movim has joined #nixos
pie__ has joined #nixos
ym555 has joined #nixos
<pie__>
anyone been experiencing wifi breakage recently
clacke_movim has left #nixos [#nixos]
obfusk has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
obfusk has quit [Client Quit]
obfusk has joined #nixos
<glenndavy>
hi all, looking for some help with whats probably a simple thing; I can do: `nix-env -i ‘bash-interactive-5.0-p7’ and install an interactive bash shell into my profile.
<{^_^}>
[nixpkgs] @thoughtpolice merged pull request #62651 → websocat: add support for wss:// → https://git.io/fjuqk
<{^_^}>
[nixpkgs] @thoughtpolice pushed commit from @volth to master « websocat: add support for wss:// »: https://git.io/fjuzK
<glenndavy>
i need to acheive the same net effect in nix-shell and can’t workout the correct incantation to insert into my array of buildInputs to make this happen.
<{^_^}>
[nixpkgs] @michaelpj opened pull request #62707 → asciidoctor: update to 2.0.10 → https://git.io/fjuzd
clacke_movim has joined #nixos
abathur has joined #nixos
<glenndavy>
gchritensen sorry to bug you again, but bashInteractive* (afaict) doesn’t install any binaries ?
<glenndavy>
hmm, though the derivation does specfify output binaries.. i’ll look deeper firsr
<gchristensen>
glenndavy: what are you seeing to think that?
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<gchristensen>
fwiw, bashInteractive should already be in your nix-shell
<gchristensen>
since Nix uses bashInteractive as its shell
<glenndavy>
oh.. then i shouldn’t have needed to be doing this, ok
<glenndavy>
so at the end of my shell.nix i build up a PATH from the build inputs
<glenndavy>
which generally seems to work but perhaps it isn’t doing it correctly for bashInterative
<clever>
glenndavy: the buildInputs should be added to PATH automatically
clacke_movim has left #nixos [#nixos]
<glenndavy>
*blink*
<glenndavy>
has that always been the case?
<clever>
yes
<gchristensen>
yea
<clever>
at least since i started using nixpkgs (2+ years ago)
<clever>
the source shows signs of an ancient time when buildInputs wasnt a thing
<glenndavy>
:/ dang, i guess thats the danger of using googled blog posts
<gchristensen>
been that way since before january 2016
<glenndavy>
i did this a year or so back and it was only way i could get a useful shell
<glenndavy>
ok
<gchristensen>
clever: and you were here on the day I first learned of NixOS, so your 2+ years is stale :P
<glenndavy>
imagonna remove my path building/mangling and see if that magically thixes all the things
<clever>
gchristensen: dang!, i thought you where here before me!
<clever>
$ head ~/irclogs/freenode/#nixos.log
<clever>
--- Log opened Thu Jun 25 11:13:00 2015
<clever>
2016-01-15 20:56:56-!- gchristensen [~gchristen@unaffiliated/grahamc] has joined #nixos
<gchristensen>
yeah, my first join was 2016-01-15 19:56:56 (UTC)
<gchristensen>
(maybe not UTC..)
<clever>
yeah, i beat you by about 6 months
<clever>
2016-01-15 21:11:11< gchristensen> henrytill: <turning a joking conversation serious> has there been any looking / work at an embedded nixos?
<clever>
gchristensen: still need an answer? :P
fendor_ has joined #nixos
<glenndavy>
LOL faark.. ok, thanks gchristensen and clever… _not_ explicitly building my PATH has fixed everything
<glenndavy>
thansk again
<{^_^}>
[nixpkgs] @ktor opened pull request #62708 → bcompare: init at 4.2.10.23938 → https://git.io/fjuzh
<gchristensen>
yay :)
<gchristensen>
clever: not any longer :)
sestrella has joined #nixos
<glenndavy>
would love to understand why `nix search` doesn’t work. my pkg source is a local checkout.. is there a config that should point search there?
<clever>
glenndavy: nix search -f . things
alp has quit [Ping timeout: 244 seconds]
<clever>
that will load . rather then <nixpkgs>
erudition has joined #nixos
<glenndavy>
ooohh.. that worked. thanks!
<clever>
glenndavy: can you pastebin the shell.nix file?
<glenndavy>
thanks, trying now.. though, you know i fixed my problem commenting out the path building ?
<clever>
glenndavy: 80% of the things in your buildInputs wherent needed
alp has joined #nixos
<clever>
and more could likely be removed, but would depend on what your doing exactly
<glenndavy>
got it, thanks.. yep minimal is good
fendor_ has joined #nixos
<DigitalKiwi>
how can i make bluetooth auto reconnect? my mouse is such that if i turn it off/on it disconnects bluetoothctl -> connect uuid usually grabs it but that's annoying and no mouse until then, recently I found I can open blueman-assistant and hit enter a few times aand that'll connect it too...
clacke_movim has joined #nixos
o1lo01ol1o has joined #nixos
<glenndavy>
clever, seems to work, i guess if anything is now missing, i’ll just discover and re-add, but +1 thanks
clacke_movim has left #nixos [#nixos]
Anton-Latukha has quit [Quit: Leaving.]
Elon_Satoshi has quit [Read error: Connection reset by peer]
fendor_ has quit [Ping timeout: 248 seconds]
ambro718 has quit [Quit: Konversation terminated!]
clacke_movim has joined #nixos
Elon_Satoshi has joined #nixos
clacke_movim has left #nixos [#nixos]
MmeQuignon has quit [Ping timeout: 248 seconds]
clacke_movim has joined #nixos
daGrevis has joined #nixos
simon_weber has quit [Quit: Connection closed for inactivity]
goibhniu1 has quit [Ping timeout: 258 seconds]
clacke_movim has left #nixos [#nixos]
o1lo01ol1o has quit [Remote host closed the connection]
fenedor has quit [Quit: Leaving]
MmeQuignon has joined #nixos
emptyflask has joined #nixos
<{^_^}>
[nixpkgs] @danbst merged pull request #53991 → build-support/make-initrd: Don't derive derivation name from file name → https://git.io/fhCrY