filterfish_ has quit [(Ping timeout: 258 seconds)]
acertain has quit [(Ping timeout: 258 seconds)]
zraexy has quit [(Quit: Leaving.)]
CustosLim3n has quit [(Excess Flood)]
propumpkin has quit [(Quit: My MacBook Pro has gone to sleep. ZZZzzz…)]
CustosLimen has joined #nixos
zraexy has joined #nixos
hydraz has joined #nixos
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
indi_ has quit [(Ping timeout: 258 seconds)]
acowley_away is now known as acowley
takle has joined #nixos
simukis has quit [(Ping timeout: 255 seconds)]
ertes has quit [(Ping timeout: 240 seconds)]
[0x4A6F] has quit [(Ping timeout: 246 seconds)]
ma9e has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
[0x4A6F] has joined #nixos
newhoggy has quit [(Remote host closed the connection)]
filterfish has joined #nixos
filterfish has quit [(Remote host closed the connection)]
filterfish has joined #nixos
newhoggy has joined #nixos
takle has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
filterfish has quit [(Ping timeout: 260 seconds)]
mudri has quit [(Quit: WeeChat 1.7.1)]
acowley is now known as acowley_away
pie_ has quit [(Ping timeout: 240 seconds)]
acertain has joined #nixos
eacameron has joined #nixos
newhoggy has quit [(Remote host closed the connection)]
eacameron has quit [(Ping timeout: 252 seconds)]
indi_ has joined #nixos
newhoggy has joined #nixos
fnljk has quit [(Quit: brb/bbl)]
indi_ has quit [(Client Quit)]
newhoggy has quit [(Ping timeout: 240 seconds)]
eacameron has joined #nixos
acowley_away is now known as acowley
georges-duperon has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Ping timeout: 260 seconds)]
deepwater has joined #nixos
newhoggy has joined #nixos
ertes has joined #nixos
byteflame has joined #nixos
newhoggy has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
Tipping_Fedora has quit [(Ping timeout: 268 seconds)]
newhoggy has quit [(Remote host closed the connection)]
johnramsden has joined #nixos
mbrgm has quit [(Ping timeout: 260 seconds)]
mbrgm has joined #nixos
derjohn_mob has quit [(Ping timeout: 260 seconds)]
Supersonic112 has quit [(Disconnected by services)]
Supersonic112_ has joined #nixos
Tipping_Fedora has joined #nixos
Supersonic112_ is now known as Supersonic112
<johnramsden>
If I wanted to add a new option to a package, what's the normal way to do this? I looked at overrideAttrs and overlays but it wasn't clear to me if they were just customizing existing options. Or if they could be used to add new ones.
eacameron has joined #nixos
aanderse_ has joined #nixos
<aanderse_>
brand new to nixos trying to figure out how i can access the "qmake" command if i want to do some light development
<aanderse_>
been searching and can't find much reference...
<johnramsden>
aandersee: You probably need to install qt
<aanderse_>
yes it is installed
<aanderse_>
i did a find /nixos -name qmake and found qmake
<aanderse_>
but i'm thinking that there is a better way than that
eacameron has quit [(Ping timeout: 240 seconds)]
<johnramsden>
What do you mean? Wouldn't you use it just like you would anywhere else?
<aanderse_>
well for example when i told my configuration file to install gcc i can call gcc from any regular console
<aanderse_>
just type gcc and good to go
<aanderse_>
but i can't just type qmake
<johnramsden>
You might need to open a new console to get it added to your path, or reboot. That's just a guess though.
<johnramsden>
I can just type qmake
<johnramsden>
and it works
<aanderse_>
can you share parts of your configuration file with me perhaps?
takle has joined #nixos
<NixOS_GitHub>
[nixpkgs] acowley opened pull request #25307: llvm: Add the LLVM OpenMP components (master...llvm-openmp) https://git.io/v9CJe
<tilpner>
aanderse_ - If you just need it quickly you can run $ nix-shell -p qt5Full and have it available
<NixOS_GitHub>
[nixpkgs] acowley opened pull request #25308: glfw: fix the install name id on darwin (master...glfw-id) https://git.io/v9CJJ
<tilpner>
aanderse_ - Otherwise, you probably want to add it to your environment with nix-env or systemPackages
derjohn_mob has joined #nixos
<johnramsden>
anderse: My nix config? I haven't added anything qt related
<aanderse_>
hmm ok...
<aanderse_>
i just typed "nix-shell -p qt5Full" and a bunch of packages are downloading now
<NixOS_GitHub>
[nixpkgs] acowley opened pull request #25309: rtags: link to libclang.dylib at absolute path on darwin (master...rtags-id) https://git.io/v9CJt
<tilpner>
I don't know enough about qt to tell if a smaller package (not full) would have been sufficient for whatever you're doing
<johnramsden>
aanderse I added all my packages through environment.systemPackages so if you added yours a different way maybe that's why.
takle has quit [(Ping timeout: 255 seconds)]
<aanderse_>
johnramsden: i only installed via systemPackages as well
<aanderse_>
but i did not explicitly list qt
byteflame has quit [(Ping timeout: 240 seconds)]
sellout- has joined #nixos
feddasch has joined #nixos
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 268 seconds)]
hyphon81 has joined #nixos
contrapumpkin has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 2 new commits to master: https://git.io/v9CJo
<NixOS_GitHub>
nixpkgs/master 32ae4bf Dan Peebles: stdenv-generic: add meta attribute checking...
<NixOS_GitHub>
nixpkgs/master 76137a8 Daniel Peebles: Merge pull request #25304 from copumpkin/check-meta...
vaibhavsagar has joined #nixos
gchristensen has quit [(Quit: WeeChat 0.4.2)]
filterfish has joined #nixos
cpennington has joined #nixos
byteflame has joined #nixos
zeus_ has quit [(Remote host closed the connection)]
Jookia has joined #nixos
<Jookia>
Has anyone gotten SELinux going in NixOS? How hard is it to do?
byteflame has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
aanderse_ has quit [(Quit: Konversation terminated!)]
newhoggy has quit [(Remote host closed the connection)]
filterfish has quit [(Ping timeout: 260 seconds)]
takle has joined #nixos
takle has quit [(Ping timeout: 268 seconds)]
thc202 has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
hexagoxel has quit [(Ping timeout: 260 seconds)]
filterfish has joined #nixos
hyphon81 has quit [(Ping timeout: 268 seconds)]
filterfish has quit [(Read error: Connection reset by peer)]
newhoggy has quit [(Remote host closed the connection)]
filterfish has joined #nixos
newhoggy has joined #nixos
acowley is now known as acowley_away
hexagoxel has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
Infinisil has quit [(Ping timeout: 260 seconds)]
roconnor has joined #nixos
filterfish has quit [(Ping timeout: 260 seconds)]
matthewbauer has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/v9CTu
<NixOS_GitHub>
nixpkgs/master 1a4ca22 Dan Peebles: treewide: fix assorted issues revealed by the meta checker...
dfranke has quit [(Ping timeout: 245 seconds)]
byteflame has joined #nixos
nicknovi1 has joined #nixos
byteflame has quit [(Ping timeout: 240 seconds)]
ericsagnes has quit [(Ping timeout: 260 seconds)]
MinceR has quit [(Ping timeout: 258 seconds)]
<NixOS_GitHub>
[nixpkgs] jerith666 opened pull request #25310: fix 'command-not-found: is a directory' error (master...command-not-found-is-a-dir) https://git.io/v9CTX
takle has joined #nixos
Wizek_ has joined #nixos
dfranke has joined #nixos
MinceR has joined #nixos
takle has quit [(Ping timeout: 260 seconds)]
schoppenhauer has quit [(Ping timeout: 240 seconds)]
nicknovi1 has quit [(Ping timeout: 240 seconds)]
Infinisil has joined #nixos
schoppenhauer has joined #nixos
batdog has left #nixos ["WeeChat 1.7"]
<NixOS_GitHub>
[nixpkgs] peterhoeg opened pull request #25311: display-manager: systemd-udev-settle serves no purpose, boot 1.5s faster (master...f/udev-settle) https://git.io/v9CkZ
vaibhavsagar has quit [(Ping timeout: 260 seconds)]
systemfault has joined #nixos
systemfault has quit [(Max SendQ exceeded)]
systemfault has joined #nixos
anton____ has quit [(Ping timeout: 258 seconds)]
newhoggy has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
newhoggy has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
anton___ has joined #nixos
ericsagnes has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/v9CIB
<NixOS_GitHub>
nixpkgs/master 90b9719 Dan Peebles: treewide: fix the remaining issues with meta attributes
roconnor has quit [(Ping timeout: 240 seconds)]
cpennington has quit [(Ping timeout: 240 seconds)]
Wizek_ has quit [(Ping timeout: 252 seconds)]
newhoggy has joined #nixos
vaibhavsagar has joined #nixos
matthewbauer has quit [(Remote host closed the connection)]
matthewbauer has joined #nixos
Ivanych has joined #nixos
<jluttine>
why isn't pythonPackage.Mako named pythonPackage.mako (lower case initial letter)?
<jluttine>
i find it confusing
<jluttine>
i'd fix that if it's ok. but it might break some people's configs
<jluttine>
and in what order are the packages in pkgs/top-level/python-packages.nix ? sometimes they are in a bit like alphabetical order but then again not at all
magnetophon1 has quit [(Remote host closed the connection)]
takle has joined #nixos
<jluttine>
also pythonPackage.Logbook starts with an upper case letter :o
<johnramsden>
If I wanted to add a new option to a package, what's the normal way to do this? I looked at overrideAttrs and overlays but it wasn't clear to me if they were just customizing existing options. Or if they could be used to add new ones.
newhoggy has quit [(Remote host closed the connection)]
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
newhoggy has joined #nixos
kampfschlaefer has joined #nixos
srhb has quit [(Quit: Quit)]
newhoggy has quit [(Ping timeout: 240 seconds)]
srhb has joined #nixos
takle has joined #nixos
newhoggy has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
hotfuzz has quit [(Quit: WeeChat 1.7)]
newhoggy has quit [(Ping timeout: 240 seconds)]
ryantrinkle has joined #nixos
newhoggy has joined #nixos
newhoggy has quit [(Ping timeout: 255 seconds)]
k2s has joined #nixos
k2s has quit [(Client Quit)]
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] ericsagnes opened pull request #25318: stdenv-generic: add meta attributes checks (master...feat/stdenv-metacheck) https://git.io/v9COi
newhoggy has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
mudri has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] DavidEGrayson opened pull request #25319: cmake: Prevent it from searching for packages in / when cross-compiling for Windows (master...pr_cmake_search_path) https://git.io/v9C33
newhoggy has quit [(Ping timeout: 260 seconds)]
takle has joined #nixos
MP2E has quit [(Quit: leaving)]
takle has quit [(Ping timeout: 260 seconds)]
<NixOS_GitHub>
[nixpkgs] offlinehacker pushed 2 new commits to master: https://git.io/v9C3u
<NixOS_GitHub>
nixpkgs/master 5539f8a Victor Calvert: xpra: 2.0.1 -> 2.0.2
<ambro718>
WTF? Can someone explain? Why hardcode to remove this?
mudri has quit [(Ping timeout: 245 seconds)]
<ambro718>
it also has the issue in that it only succeeds in eradicating these files if more than two packages provide /share/mime, if only one does (shared_mime_info) then the whole /share/mime folder will be symlinked anyway
therp has quit [(Quit: Leaving)]
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/v9CcU
<NixOS_GitHub>
nixpkgs/master adadf7e Michael Raskin: androidsdk: meta.url to meta.homepage
mudri has joined #nixos
<ambro718>
how would one use a NixOS chroot in a non-NixOS environment?
<ambro718>
Suppose I installed nix on non-NixOS and I want to chroot into a proper NixOS environment that has all the right environment settings and such, where stuff is more likely to work by default
orivej has joined #nixos
<Filystyn>
i have source i need to compile welll i have sdl2 lib but cmake does nto see it
<Filystyn>
what can cause this?
<Filystyn>
do i need some SPECIAL cmake to have it work or what ever i have to do? To make it search libs so it can find it
<sphalerite>
Filystyn: use nix-shell
<sphalerite>
Filystyn: nix-shell -p cmake SDL2
<sphalerite>
It should work in that shell
molec has joined #nixos
molec is now known as meloc
meloc has quit [(Changing host)]
meloc has joined #nixos
<sphalerite>
Do we have a stock response/article for "I installed some libs but nothing's finding them" anywhere? It's one of the most common questions here...
<sphalerite>
Filystyn: cmake doesn't search the user profile or indeed the system profile for libraries. The idea is that dependencies should always be declared, so that the result of building some piece of software won't be different from one machine to the next
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/v9CcC
<NixOS_GitHub>
nixpkgs/master 316d0ff Michael Raskin: libsamplerate: update license information (in effect since 0.1.9)
<Filystyn>
ok..
<Filystyn>
ok i think i got this hehe very strange but maybe this is a good way
<Filystyn>
nto hard so i am ok with it
<Filystyn>
thank you for your help
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/v9Cc2
<NixOS_GitHub>
nixpkgs/master 4a207b1 Michael Raskin: gimpPlugins.resynthesizer: fix build
Jookia has joined #nixos
<Jookia>
Hmm, I don't think SELinux would work with /nix/store
<Filystyn>
I have little problem i cant specifie dots in line you shown sphalerite
gloaming has joined #nixos
<gloaming>
Hi :) How do I select which compiler to use for a package? I'm trying to build wxWidgets on OSX using GCC instead of Clang
<gloaming>
I tried setting CXX in overrideAttrs but Nix seems to ignore it
<Filystyn>
how to pass dot in a name ;-)
griff_ has quit [(Quit: griff_)]
<sphalerite>
Filystyn: what do you mean?
cfricke has quit [(Quit: WeeChat 1.7.1)]
<sphalerite>
gloaming: I think you'd want to override the stdenv that it uses as appropriate. I'm not sure if there is a gcc stdenv for OSX, but if you have a look using nix-repl you might find something useful
<Filystyn>
i had to pass pkg-config-0.29 but passing pkgconfig fixed it ;-)
<Filystyn>
not sure tho how to be more precise with it
<Filystyn>
last thing it seems there is no gme lib in repository ;-)
mudri has quit [(Ping timeout: 258 seconds)]
<Filystyn>
ok i got it compiled
<Filystyn>
sphalerite would it be a problem to explain me a bit more bit by bit the sens of using nix shell here and what happens
newhoggy has quit [(Remote host closed the connection)]
newhoggy has joined #nixos
griff_ has joined #nixos
<sphalerite>
Filystyn: nix-env -qaP will list the attribute paths for each package. The attribute path is the one you want to use with nix-shell
<sphalerite>
Also, pretty good timing because I added libgme to nixpkgs just last week!
osa1_ has joined #nixos
<Filystyn>
lol ;-)
<NixOS_GitHub>
[nixpkgs] volth opened pull request #25321: Reaver database on readwrite medium (master...reaver-rw-database) https://git.io/v9CCn
bennofs has quit [(Ping timeout: 246 seconds)]
osa1_ is now known as osa1
ibor has joined #nixos
osa1 has quit [(Changing host)]
osa1 has joined #nixos
DIzFer has joined #nixos
ibor has quit [(Ping timeout: 260 seconds)]
cfricke has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ma27 opened pull request #25323: programs.zsh.syntax-highlighting: support custom highlighting patterns (master...zsh/support-pattern-highlighters) https://git.io/v9CCA
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 3 new commits to master: https://git.io/v9CCp
<NixOS_GitHub>
nixpkgs/master 6e53534 Tomas Hlavaty: sbcl: 1.3.16 -> 1.3.17
<NixOS_GitHub>
nixpkgs/master f1cf812 Tomas Hlavaty: sbcl: enable threadSupport also on aarch64-linux
<NixOS_GitHub>
nixpkgs/master f44165f Michael Raskin: Merge pull request #25303 from tohl/master...
hotfuzz has joined #nixos
<Filystyn>
ok one more
<Filystyn>
Could NOT find OpenGL (missing: OPENGL_gl_LIBRARY OPENGL_INCLUDE_DIR) how to point it with nix-shell when i pass -p cmake xxx
<Filystyn>
?
noam has quit [(Ping timeout: 240 seconds)]
thc202 has joined #nixos
<Jookia>
did you include 'mesa' in your deps
<Filystyn>
you mean -p cmake mexa xxx
<Filystyn>
?
<Filystyn>
worked ;-)
<Filystyn>
ok guys one more Could NOT find PythonInterp: Found unsuitable version "2.7.13", but i pass python3 and it still complains this is not it
<Jookia>
what version does it want
<Filystyn>
.3 at leats i passed python33 and it still syas it sees 2.7
<Filystyn>
maybe it grabs older version from system ?
<Jookia>
Filystyn: in the shell, whats 'which python' say
<Filystyn>
i just need to use new one.... God so irritating ;>
<gloaming>
sphalerite: Thanks! Seems to be on the right track. Sadly configure dies with "gcc-wrapper-5.4.0/bin/as: assembler (clang) not installed"
<gloaming>
Does this mean clang needs to be specified as a dependency somehow?
Guest98689 has quit [(Ping timeout: 260 seconds)]
noam has joined #nixos
<Filystyn>
just can't force it to use pyt3
<Filystyn>
can i remove dependence when I am in the nix shell ?
<hyper_ch>
hmmm, is this bad? syswrite() is deprecated on :utf8 handles at /nix/store/d51apm4c8mr5hiwh134vdkzgivf4hrl7-perl-File-Slurp-9999.19/lib/perl5/site_perl/5.24.1/File/Slurp.pm line 506.
meloc has quit [(Quit: Connection closed for inactivity)]
gloaming has quit [(Ping timeout: 260 seconds)]
georges-duperon has joined #nixos
Infinisil has quit [(Ping timeout: 245 seconds)]
cpennington has joined #nixos
cpennington has quit [(Remote host closed the connection)]
mbdb92 is now known as nithor
takle has quit [(Remote host closed the connection)]
ericsagnes has joined #nixos
alx741 has joined #nixos
takle has joined #nixos
<hyper_ch>
hmmm, getting syswrite() is deprecated on :utf8 handles at /nix/store/d51apm4c8mr5hiwh134vdkzgivf4hrl7-perl-File-Slurp-9999.19/lib/perl5/site_perl/5.24.1/File/Slurp.pm line 506.
osa1 has quit [(Ping timeout: 240 seconds)]
derjohn_mob has joined #nixos
<DIzFer>
same
<DIzFer>
everything works fine though
<hyper_ch>
question is now: Is that bad and are we all gonna die?
<Jookia>
hyper_ch: are you installing nixos on your pacemaker
<hyper_ch>
isn't everyone?
takle has quit [(Ping timeout: 268 seconds)]
<Jookia>
nope
<hyper_ch>
because you can't afford a pacemaker?
sellout- has quit [(Quit: Leaving.)]
nicknovi1 has joined #nixos
<MichaelRaskin>
Some people are cordially privileged enough to be able to survive up to 4 hours without electronics, and even longer if asleep!
<hyper_ch>
BLASPHEMY!!!
newhoggy has quit [(Remote host closed the connection)]
<MichaelRaskin>
Well, yes, truth often does turn out to be blasphemous.
pie_ has joined #nixos
<hyper_ch>
so, MichaelRaskin, you know why the above ouput appears on trying to upgrade=
pie_ has quit [(Changing host)]
pie_ has joined #nixos
takle has joined #nixos
takle has quit [(Ping timeout: 258 seconds)]
<MichaelRaskin>
Well, it doesn't print that for me, but I am not using literaly NixOS
eacameron has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/v9Czt
<NixOS_GitHub>
nixpkgs/master f1c7d5a Michael Raskin: gimpPlugins.resynthesizer2: fix build
MichaelRaskin has quit [(Remote host closed the connection)]
takle has joined #nixos
Wizek_ has quit [(Ping timeout: 252 seconds)]
gchristensen has joined #nixos
takle has quit [(Ping timeout: 260 seconds)]
byteflame has joined #nixos
takle has joined #nixos
mekeor has joined #nixos
mekeor has quit [(Client Quit)]
newhoggy has joined #nixos
Itkovian has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
tvon has joined #nixos
tvon_ has joined #nixos
tvon has quit [(Read error: Connection reset by peer)]
<gchristensen>
I have no idea how to fix that :P you might try asking in #emacs?
nh2 has joined #nixos
<mekeor>
true
Filystyn has quit [(Read error: Connection reset by peer)]
georges-duperon has quit [(Ping timeout: 240 seconds)]
jsgrant has quit [(Ping timeout: 260 seconds)]
darlan has joined #nixos
georges-duperon has joined #nixos
newhoggy has joined #nixos
georges-duperon has quit [(Client Quit)]
<Oo[m]>
Hi, is it possible for a package to read a file only readable by root ? I would like to copy /sys/firmware/acpi/tables/DSDT (I'm trying to write a nix package to fix a faulty ACPI table.) If it is impossible I will copy it manually but it would be better to do it inside the .nix file, because if the file changes after a BIOS update, I would only need to use nixos-rebuild
jsgrant has joined #nixos
hotfuzz_ is now known as hotfuzz
newhoggy has quit [(Ping timeout: 255 seconds)]
<gchristensen>
Oo[m]: do you mean when building the package?
<gchristensen>
or after the expression has built
<Oo[m]>
when building the package
<gchristensen>
mmm that seems strange. can you tell me why you want to copy the file when building?
bennofs has joined #nixos
<gchristensen>
does /sys/firmware/acpi/tables/DSDT become corrupted later on?
<gchristensen>
so you copy the file, patch it, and write it back?
<joachifm>
gchristensen: you can tell linux to use a different table
pie_ has joined #nixos
<gchristensen>
joachifm: I'm trying to extract info from Oo[m] so I can lead them to a better solution :)
<joachifm>
I think it makes more sense to create a hardware-specific package and just copy the dsdt file instead of trying to read it at build time
<Oo[m]>
I make a cpio archive and then use boot.initrd.prepend to load it
dridus has quit [(Quit: znc? wassat?)]
<gchristensen>
ahhh I see, that is tough...
<jophish>
gchristensen: regarding that paper on scheduling jobs. I think it should be possible to do better with Hydra/Hercules
<gchristensen>
oh, jophish?
<jophish>
Because one should be able to calculate some kind of estimate on how long a job will take to build
newhoggy has joined #nixos
<gchristensen>
based on prior knowledge?
<jophish>
I imagine that the next revision of pkgs.firefox will take a similar amount of time to compile as the current revision
<jophish>
gchristensen: exactly
<gchristensen>
true
<gchristensen>
I was thinking on the plane last night, will the removal of the remote build hook make it harder to implement interesting schedulers? (cc niksnut)
<jophish>
gchristensen: the libstore api approach should make it possible to reclaim that behaviour
Itkovian_ has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<jophish>
This prints out the path whenever a derivation needs to be built
<jophish>
there's no reason it couldn't farm that work out to a remote builder instead of using the underlying store implementation
<jophish>
in fact, that's exactly the intention with hercules
Filystyn has joined #nixos
<gchristensen>
interesting
Filystyn has quit [(Changing host)]
Filystyn has joined #nixos
newhoggy has quit [(Ping timeout: 260 seconds)]
<Filystyn>
can someone explain me how does nix-shell -p cmake dependencies work when i leave it where are the so libs stored ? or just give me link
<pie_>
clever, i never did hear what you think about the suckless thing iirc?
<gchristensen>
jophish: `std::for_each(paths.begin(), paths.end(),` will this guarantee correct dependency ordering?
<Oo[m]>
joachifm: yes, it seems I have no other choice than to copy the DSDT table, that's not a big deal :) , thanks for the quick answers !
<jophish>
gchristensen: it's only printing them out!
<gchristensen>
jophish: I know lol but if I copypasta that code will it do the right thing? :P
<jophish>
gchristensen: well, you're going to have to put those paths in the store at some point. And assuming you don't do that in a naughty way (i.e. mount the store as writable yourself) then the daemon will make sure the store invariant is held
<gchristensen>
ok
<jophish>
if that's not the case then I suppose you'll have to do some toposorting using the output of nix-store --query --something
<joachifm>
Oo[m]: if you care to, you could look into adding what you come up with to the nixos-hardware repo
* gchristensen
's just trying to find the right thing to do
<jophish>
gchristensen: if PathSet is a std::set underneath, then it will be a bit more complicated!
<gchristensen>
:)
newhoggy has joined #nixos
ryantrinkle has quit [(Ping timeout: 260 seconds)]
<gchristensen>
the next obvious question is would anyone like to make a Rust crate for these? :P
<Oo[m]>
joachifm: if I finally make it work, I'm a complete beginner with NixOS :)
<jophish>
gchristensen: a rust crate for the libstore api?
<bennofs>
gchristensen, jophish: nix apis <-> rust bindings would be awesome
<gchristensen>
for all the APIs needed to make a store implementation
<jophish>
Haskell too, please!
<jophish>
The plan for my store api implementation was just to forward all calls over some socket, to be responded to by a local haskell server
<jophish>
get it away from C++ as quickly as possible
newhoggy has quit [(Ping timeout: 245 seconds)]
<gchristensen>
so like
<gchristensen>
you want the build script hook :P
endformationage has joined #nixos
newhoggy has joined #nixos
sellout- has joined #nixos
<jophish>
gchristensen: yes! but this is even better, as it gets called whether the store paths exists or not
<jophish>
where as the build script hook only gets called if the path needs to be built
<bennofs>
jophish: does it get called on the drv or on the output of the drv?
<jophish>
bennofs: the drv
<bennofs>
oh nice
<jophish>
let me add some example output to the readme
roconnor has joined #nixos
<gchristensen>
I mean I get it, but it seems you want the build script hook :P
derjohn_mob has quit [(Ping timeout: 246 seconds)]
<NixOS_GitHub>
[nixpkgs] DavidEGrayson opened pull request #25330: cmake: Prevent it from looking for packages in / when cross-compiling for Windows. (staging...pr_staging_cmake) https://git.io/v9Crz
<NixOS_GitHub>
[nixpkgs] primeos pushed 1 new commit to master: https://git.io/v9Crp
<NixOS_GitHub>
nixpkgs/master 8528136 Michael Weiss: desktop-managers: Use a black BG as fallback...
systemfault has joined #nixos
<NixOS_GitHub>
[nixpkgs] primeos closed pull request #25101: desktop-managers: Use a black BG as fallback (master...desktop-manager-background-fix) https://git.io/v9UzB
orivej has quit [(Ping timeout: 252 seconds)]
reinzelmann has joined #nixos
jsgrant has quit [(Quit: Peace Peeps. o/ If you need me asap, message me at msg[at]jsgrant.io & I'll try to get back to you within 24 hours.)]
newhoggy has joined #nixos
osa1 has joined #nixos
seanparsons has quit [(Quit: WeeChat 1.7.1)]
newhoggy has quit [(Ping timeout: 260 seconds)]
khebbie has joined #nixos
xadi has quit [(Quit: Leaving.)]
xadi has joined #nixos
Guest98689 has joined #nixos
<simpson>
What I like about Nix today: If I rebase, but only to do fixups and patch reordering, then nix-build fully caches the build since the source files haven't changed.
<simpson>
Better than Make.
peacememories has joined #nixos
newhoggy has joined #nixos
<clever>
:D
roconnor has quit [(Ping timeout: 268 seconds)]
<clever>
you just need to filter out .git when hashing it
<clever>
simpson: ccache gives you similiar benefits at a .o level
<clever>
simpson: and i have had thoughts on how to recreate ccache within nix, in a pure manner
s33se_ has joined #nixos
<simpson>
clever: We have similar plans in Monte, but using Nix directly since we don't have anything between us and Nix.
<clever>
simpson: my idea was to ${./foo.c} every source and buildenv a bunch of ./foo.h's together, so it depends on the hashes of the individual source files, rather then the tar it came from
newhoggy has quit [(Ping timeout: 260 seconds)]
<clever>
but you now need to rewrite every makefile in nix
roconnor has joined #nixos
<simpson>
And your store gets filled with tiny files. Maybe that's okay?
<clever>
its no worse then /nix/store/.links/
<bennofs>
clever: one problem is that you'd have to copy every single file before building
<clever>
yeah
<clever>
but also, you get distcc for free
<clever>
send 5 .c files to every build slave in the hydra cluster
<clever>
then gather all the .o's up and link
<Dezgeg>
solving incremental building is the biggest problem I guess (since you don't know a priori whether foo.c will include bar.h or not)
<clever>
Dezgeg: it scans the #include statements, and spits out a partial Makefile rule
<clever>
yeah
<clever>
but in the best case, you can gcc -M every file once, then reuse the .o from the binary cache and skip the compile
<clever>
or, because of the purity of nix, you can gcc -M once, to create a dep-tree for the entire project
<clever>
and just lock that in at commit time
<clever>
and if somebody ever tries to violate it, the sandbox will cause it to fail
<clever>
thats one thing ive messed up sometimes with hand-written Makefile's
<clever>
if you #include something new and forget to update the makefile, weird things happen
takle has quit [(Remote host closed the connection)]
<Dezgeg>
well no one really writes header dependencies manually these days
<clever>
i do when the project is transitioning from a single gcc command to something slightly more complx
<clever>
and i dont want to bother with cmake or similiar
<Dezgeg>
for that you can copy paste a minimal 100-line makefile from somewhere
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<bennofs>
or copy-paste a minimal 3-line CMakeLists from somewhere :p
<Dezgeg>
or that :P
peacememories has joined #nixos
<ahmedtd>
How can I find any failures that might result from a particular commit to nixpkgs master?
_Vi has quit [(Ping timeout: 245 seconds)]
newhoggy has joined #nixos
peacememories has quit [(Client Quit)]
ryantrinkle has joined #nixos
simukis has quit [(Quit: simukis)]
newhoggy has quit [(Ping timeout: 252 seconds)]
andymandias has quit [(Ping timeout: 255 seconds)]
<sphalerite>
NickHu: are you using the android tools from the android site?
<sphalerite>
ahmedtd: using nox, I think
andymandias has joined #nixos
alunduil has quit [(Quit: leaving)]
<ahmedtd>
sphalerite: I was looking for a way to see it on hydra --- the commit in question is a mass-rebuild, so `nox-review` keeps failing on random already-broken packages
<ahmedtd>
I think I've found how to navigate hydra well enough to get most of what I want
<nh2>
clever: I find the build-system/nix integration a very interesting topic. For C, your approach might work just fine, but I suspect in the long term we will see that all build systems become fully reproducible and resumable; then nix should have support for using them somehow
<sphalerite>
oh right
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 2 new commits to staging: https://git.io/v9Cin
<NixOS_GitHub>
nixpkgs/staging 934ed02 David Grayson: cmake: Prevent it from looking for packages in /...
<NixOS_GitHub>
nixpkgs/staging 8e3ed32 Michael Raskin: Merge pull request #25330 from DavidEGrayson/pr_staging_cmake...
<sphalerite>
nh2nh2: that sounds very optimisctic
<clever>
nh2: it would need not just checkpointing and the ability to resume, but also the ability to do parts of the build in parallel and pure, and then merge the results up
<nh2>
sphalerite: I don't think it's optimism, just that it's the only really practical way to go as things grow. Look at Google and Facebook, they put significant man-time effort into making fully reproducible resumable builds because otherwise they simply cannot manage the task
<sphalerite>
wlel, by that logic nix will take over the world in the long term :D
<nh2>
sphalerite: right!
simukis has joined #nixos
bennofs has quit [(Ping timeout: 240 seconds)]
<nh2>
clever: parallel and pure I get, but the "merge results up" I don't quite get, what do you mean?
<clever>
nh2: things like linking the .o's from each parallel process into the final program
ertes has quit [(Quit: Bye!)]
<sphalerite>
wlel, by that logic nix will take over the world in the long term :D
takle has joined #nixos
ertes has joined #nixos
newhoggy has joined #nixos
<nh2>
ah, you are thinking about a case as in C, where it's common to have multiple compiler invocations, yes. What I had in mind is that to nix, the concept of object files might become unimportant once we have build systems (one might rather need to say "toolchains", as you can write whatever smart build system if e.g. gcc tricks you into impurity) that always build reproducibly, given any incremental snapshot of their state. Nix would have to trust thes
<ibor>
Does anybody have a good pointer on how to do Ocaml development in NixOS?
dridus has joined #nixos
<jophish>
NickHu: SonarPulse is, however he's on vacation for a couple of weeks at the moment
<jophish>
You could ask ryantrinkle
takle has quit [(Ping timeout: 255 seconds)]
<jophish>
clever: I think that for that system to really work well we need an intensional store
<jophish>
Add a comment to a file, that gets recompiled and the rest of the build is cached
newhoggy has quit [(Ping timeout: 252 seconds)]
<nh2>
clever: my favourite world state would be something similar to this: you'd declare `src` in nix as usual, and the build system can produce, or resume from, a "build" directory when building (like the common `build/` dir in cmake, or `dist/` for Haskell). Any build system that is "nix-certified" (:D) would have to make the promise that no matter what is in that build directory, it will generate the same output as if no build directory was provided for
<clever>
nh2: i was thinking more along the lines of hashing every input to each .o file, to enforce that without the build system having to make a promise
nicknovi1 has joined #nixos
leat has quit [(Read error: No route to host)]
<nh2>
clever: for C that works due to its simplicity, but for other systems that is close to impossible because the only component that is smart enough to know what all the inputs are is the compiler (in other words, there is no equivalent to `gcc -M`). Even in C it is already the case a bit that the compiler discovers only at run-time which things are inputs: `gcc -M` will tell you much of the inputs, but not all of them; e.g. it may not include all of th
<clever>
nh2: yeah, and cases like haskell and java are a messy area
<clever>
nh2: since they lack proper header files
<simpson>
Yeah, what's the point of having compilation units if you can't actually compile them independently!?
<clever>
nh2: i believe there is a ghc -M, but you would need to get the .hi out of the .hs, and then divorce the hi from it, so changes to the .hs dont cascade and cause rebuilds
<clever>
and then you need a way to rebuild the .hi quickly, to check if it has changed
<clever>
otherwise, it wont notice the changes
nicknovi1 has quit [(Ping timeout: 255 seconds)]
<nh2>
simpson: it is always a trade-off. Simple file organisation is already a merit by itself, but independent compilation / the header file approach will always be in conflict with going for best performance (for example, C's approach is great for independent compilation and thus very parallel builds, but e.g. Haskell's approach gives you reliable cross-module inlining and thus much better performance than would be possible without, or would be possible
<clever>
nh2: what if haskell just included the AST in the .o's and did more inlining at link time?
newhoggy has joined #nixos
<clever>
nh2: for polymorphic types like foo :: [a] -> [a], it cant know what functions to call at compile time, and has to either use sucky function pointers, or defer the compile until later on
<slyfox>
in C world -flto could arguably do cross-module inlining as well
<jophish>
clever: by link time it's too late for many optimisations, unless GHC starts to use a custom linker
<simpson>
nh2: Monte's answer is to strictly compile modules as compilation units individually, but we are currently building a "muffin" mode where the late-binding of modules can instead happen early and then machine code can be generated.
<simpson>
Because working modules are more important than speed.
<clever>
jophish: then how would haskell deal with foo :: [a] -> [a] in a library?, would the performance always suck, compared to if the same code was in the project?
<jophish>
clever: the .hi file is provided for libraries
<nh2>
clever: for what you said before: yes, in fact I got these thoughts when working with Haskell. As you say, there is `ghc -M`, but you simply can't rely on it. It may not include C .h files that you #include with CPP if you use it, it may not include global system deps, it may not even know ahead of time *which* .hi files a given .hs file depends on, as that depends on whether ghc decides that a function that you call in your .hs file was deemed impor
<jophish>
usually
<clever>
jophish: and would the .hi contain just the types, or the AST of the function for future optimization?
<simpson>
clever, jophish: It would be possible if GHC were structured differently, but it's a hard structure to build; it requires more self-application powers in order to get better specialization.
<jophish>
clever: if the function is deemed small enough to inline then the core will be in there
<jophish>
or everything will be in there if you have a flag such as -fexpose-all-unfoldings
<clever>
jophish: ah, so the .hi is more then just type info, and that could cause problems if i wanted to rely on the hash of the .hi to trigger future rebuilds
newhoggy has quit [(Ping timeout: 260 seconds)]
<jophish>
yeah, that's how I understand it
engblom has quit [(Read error: Connection reset by peer)]
<clever>
jophish: but allowing those rebuilds may also improve perf
<nh2>
clever jophish: that is correct
<jophish>
I'd really like to improve the build system we have at work
engblom has joined #nixos
<jophish>
we have a large source tree which at the moment is split into many large modules
<slyfox>
ghc allows you easily peek into .hi files: 'ghc --show-iface /usr/lib64/ghc-8.0.2/ghc-prim-0.5.0.0/GHC/Classes.hi'
<clever>
slyfox: ooo
<jophish>
err, many haskell packages*
<jophish>
It would be really nice if we could instead have a large soup of haskell modules, which are cached appropriately by nix
<clever>
slyfox: dang!, 8000 lines for a 31 line data definition, with zero imports
<nh2>
the general problem here is: How much smartness do you put into the compiler, how much smartness do you put into the external system calling it. I've worked on things like ghc-parmake, that try to parse `ghc -M`output and make smart decisions based on interface file contents; you quickly have to incorporate lots of the compiler's own logic, and you start feeling like "what I'm doing here is really a compiler job"
<clever>
over 100x increase in size, though i am derivating from Eq, Show, and Generic
pie_ has joined #nixos
newhoggy has joined #nixos
<clever>
nh2: i would just import GHC at that point, and tell it to parse things for me, but skip the compilg
leat has joined #nixos
nicknovi1 has joined #nixos
<nh2>
clever: right, but then you can make the argument: You are already fully trusting ghc here. So you might just as well trust it to do deterministic pure compilation giving a build cache folder
<clever>
slyfox: i am seeing a number of hashes in the .hi file, those are probably there to prevent mistakes and force rebuilds at the right time
<slyfox>
parsing template haskell is tricky without compilation
marsel has joined #nixos
<slyfox>
yes, .hi files contain ABI hashes of other .hi depends
<clever>
nh2: i was thinking something along the lines of run code X on all source files to produce state1, then hash(state1 + code Y) = compile(code Y, state1)
<slyfox>
it's how subsequent 'ghc --make' calls in already built tree try to detect and recompile things
<clever>
nh2: and if the compiler puts too much data about the neighboring files into state1, then it just recompiles the Y unit more often
<slyfox>
but those hashes are computed before assembly code generation phase
<clever>
nh2: then use the store to save those hash=output pairs, and sandboxes to enforce it only accessing the Y.hs and state1 files
<clever>
slyfox: i have noticed that if i import an out of date project with src = ./.;, that ghc --make wont always rebuild things
<clever>
slyfox: because the timestamps of all files are jan 1st 1970
<clever>
but in that case, the final output may have the same age as every source file, so it doesnt even check things
newhoggy has quit [(Ping timeout: 260 seconds)]
<slyfox>
interesting. yeah, perhaps timestamp tricks ghc at .hi import time
<clever>
cp -r ${./src}/*.hs src/
<nh2>
clever: yes, GHC relies on timestamps. There is currently no option to make it ignore timestamps an to rely on hashes only (hopefully somebody can add that in the future)
<clever>
this was enough to clean the build and purge everything
<clever>
nh2: that almost defeats the entire point of having the hashes, all it does it prevent mistakes if the timestamps get out of sync in the other direction
<nh2>
clever: I currently trick nix into getting incremental builds working with `ghc --make` by building in the actual source directory that containts `dist/`
<clever>
nh2: i havent looked into out-of-tree builds with ghc yet, so i tend to leave a mess in the source dir
<nh2>
clever: the perspective is different: GHC fundamentally relies on accurate file system timestamps. The hashes are not there to check when compilation is *necessary*, but to check when compilation is *not necessary* -- they are there so that you can `touch` a file and (in contrast to `make`) it will still figure out that nothing has changed
<clever>
yeah
<clever>
nh2: hmmm, what if i modify the source in a .hs, but not its types, the hashes of the types shouldnt change, and it may not rebuild things that depend on those types?
<slyfox>
it would make sense to change ghc to store source hash instead (or along with) timestamp
<slyfox>
i guess shake does exactly that
<clever>
slyfox: i have also thought about adding a nix backend to shake, so it can spit out a .nix file for building each .o as its own derivation
<clever>
so instead of shake directing the entire build, it will transform the dep tree and commands into a .nix tree
<nh2>
clever: some build systems (like Shake) have an additional option to use the hashing machinery also for the concept of "file system times are not reliable", but it is not common, as `stat` is extremely fast even on a 10 GB source code base, but a hash of all inputs is not; so many real-world users would rather fix their file system (which I think makes sense) or add file-system level file checksums that can be read O(1) (ZFS has them but it doesn't e
_Vi has joined #nixos
<clever>
nh2: i know somebody that has worked on a source repo so large that the stat phase of make took something like 30 minutes
newhoggy has joined #nixos
<dash>
wrong mindset
<dash>
use a 32-bit hash and store it in the mtime
<pie_>
clever, um
<nh2>
clever: yes, I image doing `read()` on that would take even longer
<pie_>
clever, um...
<pie_>
clever, um.
<nh2>
dash: interesting approach, but sounds abusive. Seems easier to make a "proper" stat field for that
<slyfox>
dash: i think it would work for ghc as ghc does exact timestamp matching, but not for make which does '>=' compaison
<nh2>
clever: for your other question: `ghc --make` will always correctly notice when you change the source: if you change a file, the timestamp changes, and ghc will check everything that depends on that file for recompilation; it will skip it if it can tell (based on hashes) that it is not necessary
<clever>
nh2: ah
ixxie has joined #nixos
newhoggy has quit [(Ping timeout: 240 seconds)]
<nh2>
clever: this is also where Haskell has one great advantage over C in terms of building: Interface file hashes and timestamp checks are there for *all* Haskell modules, even those that may be installed in some system-level `/usr/lib` or whaterver location, so `ghc --make` will correctly rebuild if you change a global file (e.g. on Debian with a system update, of course on NixOS this doesn't make sense); the typical C build cannot usually do that
newhoggy has joined #nixos
<clever>
nh2: ah, i think cmake can do that, let me check some things
<clever>
nh2: yeah, cmake flagged every system header as a depdency of the .o's
<clever>
though under nixos, that will break for nix-shell based builds, because of the lack of timestamps
<clever>
but the paths are also different
<clever>
it would need to regen the depdency info every time the include path changes
ixxie has quit [(Ping timeout: 245 seconds)]
takle has joined #nixos
<nh2>
clever: I think the timestamps are an interesting discussion point: There are only 2 ways to solve "fast incremental builds". Either you expose the timestamps, and push the trust of reproducibility into the build system (I think it is a viable thing to do because e.g. Haskellers really really want that `ghc --make` is reproducible), or you need `stat`-level hashes
newhoggy has quit [(Ping timeout: 245 seconds)]
Guest98689 has quit [(Ping timeout: 258 seconds)]
takle has quit [(Ping timeout: 246 seconds)]
<nh2>
clever: back to the topic of where the "smarts" of building are (because I like discussing this :D): By now, many build system designers have adopted the Haskell terminology of "applicative vs monadic build systems", where applicative build systems know the full dependency graph ahead of time, and monadic ones discover them as they go along. Applicative ones are easier to parallelise and analyse, monadic ones are strictly more powerful (more expressi
gloaming has joined #nixos
<clever>
i think the msg got cut off a bit, "more powerful (more expressi"
<nh2>
ah IRC
<nh2>
... more powerful (more expressive and can produce results that are not achievable with applicative ones, such as sophisticated optimisation). `make` is applicative, `shake` and `ghc --make` are monadic.
<nh2>
an interesting case is that of Google Bazel, whose execution engine is monadic like `shake`, but whose surface language is applicative (you need to declare deps instead of discovering them)
<clever>
nh2: irc is limited to 512 bytes per message, including the comamnd and channel name
<clever>
and your msg was cut off at about 454 bytes, not counting the command/chanenl
<nh2>
I also don't trust pidigin, but it seems either way I will have to split things
<bennofs>
nh2: does applicative actually have any advantage, apart from implementation simplicy? With shake for example, I don't think you can add more dependencies for the current build, but you can add dependencies for the "next" build
<simpson>
Why would you want a monadic build system? I'm not sure I understand the purpose.
<clever>
nh2: one thing i was thinking of with the nix style, is that it forces things to be more applicative, once you turn the nix sandbox on, it cant access an impurity, and you must either manualy or automaticaly include everything
<bennofs>
simpson: it allows you to do things like generate dependencies dynamically. Like reading from a file and then depending on some target depending on what is in the file
<nh2>
simpson: "discover the #includes of a file you want to build" is a monadic step
<simpson>
bennofs: Yikes. That seems never-desirable.
<simpson>
nh2: I would consider that a bug in the module subsystem of the language in question.
<bennofs>
simpson: you actually want something like that for a lot of things though. For all sorts of module systems for example
ibor has quit [(Ping timeout: 258 seconds)]
<clever>
simpson: ive also heard that GHC can run haskell code at compile-time, to do something
<simpson>
bennofs: I work with late-binding systems, so that's not a problem.
<simpson>
clever: Yeah, for non-Simple builds, I think they're called.
<bennofs>
simpson: what if you want to auto-generate another module that you `require` in some module?
<bennofs>
simpson: and "auto-generate" is another build step
<nh2>
simpson: by that deinition all languages are bugged. I think automatic dependency generation has lots of merit because otherwise humans make mistakes. I've seen countless manually written Makefiles where somebody included a header file in the .c file but forgot to update their Makefile
<bennofs>
simpson: is there any language that does not require any information about a `#include`d module (whatever #include translates to in that lang)
<clever>
nh2: yep, done exactly that, which is something i think the nix sandbox can enforce getting right
<simpson>
bennofs: Sure, e.g. Monte has a 1:1 overlap between modules and compilation units, and all linkage is late-binding.
<simpson>
It's not hard once you realize that speed isn't the most important thing.
<bennofs>
simpson: but you do need to know what symbols a module exports if some other module requires it right?
<simpson>
nh2: Well, yes, by my estimation, many languages *are* not-very-good. We've learned a lot since the 90s.
<simpson>
bennofs: No, why would you need to know that? That's resolved later, at binding time, which is usually right before runtime.
newhoggy has joined #nixos
<bennofs>
simpson: so you don't even need to know that the module that is required exists? how can you typcheck then? or is typechecking delayed to binding time as well?
<dash>
simpson: won't something like muffin need that, though
<simpson>
dash: Yes, but it's not a special case, because our modules are just plain objects.
<dash>
bennofs: typechecking is a lifestyle choice
<simpson>
bennofs: Monte is an object-based language, or "unityped" as Haskellers like to call them.
<nh2>
clever: yes, I think sandboxes are *always* right. You would need a very sophisticated sandbox in order to catch all mistakes though: In standard nix, you have the entire `src` available, so you won't notice if your Makefile lacks an entry. (Your idea with the nix-based make would solve that though.)
<clever>
nh2: yeah
<bennofs>
simpson: oh ok :) but don't you still need to know what the superclasses of some imported class are?
fresheyeball has joined #nixos
<simpson>
bennofs: Anyway, types are largely orthogonal. Look at e.g. Java with its classloader.
<clever>
nh2: the main reason for cutting down what it can read is just to minimize what can influence the hash of $out, but also stopping other impurities is a benefit
<fresheyeball>
hello out there, the time has come to add harddrives to my first nixos box
<fresheyeball>
I'd like to do so "properly" if there is such a thing
<fresheyeball>
and have them automount on boot
<simpson>
bennofs: Yes, we would have to do that just like Java, *if* we had classes. But we don't, so it's fine.
<fresheyeball>
I was thinking I could just run `nixos-generate-config` again
<clever>
fresheyeball: if you just mount them manualy and re-run nixos-generate-config it will update hardware-configuration.nix
<fresheyeball>
is that all I need to do to get nixos to see the harddrives?
<clever>
fresheyeball: you may want to make a backup of that, and then diff it to confirm it did the right thing
<fresheyeball>
clever: I put all of those files in git
Mateon1 has quit [(Ping timeout: 260 seconds)]
<fresheyeball>
backed up for sure
<clever>
then it should be easy
<fresheyeball>
ok
<fresheyeball>
so these are internal drives
<fresheyeball>
do I need to mount them manually before running nixos-generate-config
<bennofs>
simpson: oh I don't know anything about monte. so it basically only has no user-defined types (or at least no way to declare any relationships that are of importance at compile time between them)?
<fresheyeball>
why?
aneeshusa has joined #nixos
<nh2>
bennofs: yes, applicative build systems do have advantages too. For example for parallel building. In Bazel, you know ahead of time if you should allocate 1000 machines for your build or 10000. You can give a good estimation of build times, and show a progress bar. In shake, you discover all of this as you go
<clever>
fresheyeball: yeah, nixos-generate-config just scans what is currently mounted
newhoggy has quit [(Ping timeout: 260 seconds)]
<fresheyeball>
thats kinda lame
<fresheyeball>
ok
<fresheyeball>
so then to automount on startup
<fresheyeball>
I've never used fstab before
<fresheyeball>
but it looks like that is the normal linux way
<bennofs>
nh2: ah ok, it's good for pre-allocating. I wasn't taking that into account, thank you! (was thinking only about parallel building on a single machine)
<simpson>
bennofs: Those features absolutely exist. We just don't guarantee that the compiler erases them at compile time. Do you know about Ada?
<clever>
fresheyeball: dont need fstab, just mount /dev/sdc1 /media/foo && nixos-generate-config
<fresheyeball>
is that the right way for nixos?
<fresheyeball>
oh
<bennofs>
simpson: oh, so you don't guarrante to check them at compile time
<fresheyeball>
will doing so also make the new disks mount on startup?
<clever>
fresheyeball: the new config it generates will configure everything to mount up next time you boot
takle has joined #nixos
<fresheyeball>
oh nice
<fresheyeball>
ok
<fresheyeball>
so just install the drives
<fresheyeball>
reboot
<fresheyeball>
mount
<fresheyeball>
run generate
<fresheyeball>
reboot
<simpson>
bennofs: Yep. It gives us a lot more expressive power.
<fresheyeball>
drives have automounted
<clever>
fresheyeball: and nixos-rebuild switch
<fresheyeball>
ah
<fresheyeball>
ok
<fresheyeball>
ok!
<fresheyeball>
thanks clever
<fresheyeball>
you are very helpful!
marsel has quit [(Ping timeout: 258 seconds)]
<clever>
yep :)
newhoggy has joined #nixos
fresheyeball has quit [(Quit: WeeChat 1.7.1)]
ibor has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 240 seconds)]
<clever>
slyfox: i think a large chunk of the unfolding in this .hi is generics and eq
<nh2>
clever: 30 seconds stat sounds like the Google code base, assuming a stat takes roughly 1 us and a file has roughly 1000 lines of code. So I guess if you have < 100 Mloc, stat should be OK
newhoggy has joined #nixos
<clever>
nh2: lets just say the project was large enough to need petabytes of log files
<nh2>
clever: I would imagine that in such cases, the files would typically not fit on a single machine; I guess for most such cases it makes sense to bring the up-to-dateness information into userspace
<clever>
the codebase wasnt nearly that big, but from what i heard, it was still pretty massive
newhoggy has quit [(Ping timeout: 252 seconds)]
<nh2>
one thing I learned recently about stat() is that there is no asynchronous version of it. So if you want to do it e.g. on an NFS, you need to use threads to have any form of acceptable performance
<clever>
id also expect threads to help when dealing with a raid mirror and a well-designed fs
<clever>
2 threads would let both drives seek and lookup data
<clever>
nh2: also, readdir is supposed to stat for you, in bulk, but not every FS implements it fully
<nh2>
clever: unfortunately, readdir() doesn't have all the stat() information
<clever>
yeah
<clever>
and some recent FS's omit the fields it does have, because they dont store things in the dir listing
<nh2>
clever: I find it amazing that there's still so much to do in this area -- before I started looking into it, I imagined this was solved in the 90s
<fXl>
hello guys, can we do something like this with nix ? lets say i ve installed nix on my debian machine as package manager and i had Docker installed already so can i tell nix to use that docker instead its own ?
<bennofs>
fXl: nix does not use docker
<fXl>
no i meant, can we use other host packages with nix ?
<clever>
nh2: the google at the end! lol
newhoggy has quit [(Ping timeout: 240 seconds)]
<simpson>
fXl: Not really. Nix-built packages refer into the Nix store, not into /usr.
takle has joined #nixos
<bennofs>
fXl: i don't think that is supported very well. you'd have to disable sandbox builds if you want to have them available for buildings and then override the derivations for them to just do nothing
ryanartecona has joined #nixos
<nh2>
clever: the numbers work out easily though; with 50k people, that's roughly 40k LOC per person. Sounds managable. The more interesting question is how all the open-source software in the world together (Debian) can be so small!
<fXl>
okey thanks guys
<bennofs>
nh2: does that count every package available in the repos or just the base install?
<nh2>
clever bennofs: I must be wrong, that seems to be only the base install
<nh2>
because otherwise Linux and OpenOffice together would already make most of Debian
<clever>
nh2: is the count for debian just the package manager and core os scripts?, or every package in the package manager?
matthewbauer has quit [(Remote host closed the connection)]
matthewbauer has joined #nixos
takle has joined #nixos
matthewbauer has quit [(Remote host closed the connection)]
matthewbauer has joined #nixos
newhoggy has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
systemfault has quit [(Quit: Bye!)]
aneeshusa has quit [(Quit: WeeChat 1.7)]
newhoggy has quit [(Ping timeout: 252 seconds)]
<romildo>
I have installed opensmtp, fetchmail, procmail and mutt, but there is no /var/spool/romildo directory on my NixOS system. Which package/module is supposed to handle it?
<romildo>
I mean, there is no /var/spool/mail/romildo
Mateon1 has quit [(Remote host closed the connection)]
<jsgrant_>
On a new box, am I missing an obvious way to do a "nixos-install" that will install from a version newer than the given disk? I have 17.09 set in my system version -- and set my nix-channel ... but not sure what/if anything I need to do?
Mateon1 has joined #nixos
newhoggy_ has quit [(Ping timeout: 260 seconds)]
revtintin has quit [(Quit: WeeChat 1.7.1)]
MichaelRaskin has joined #nixos
romildo has quit [(Quit: Leaving)]
* jsgrant_
has a bit of a hard-time believing, that it's not somefactor fo supported -- but can't seem to find any info on it. :^P
<clever>
jsgrant_: what exactly are you trying to change/do?
takle has quit [(Ping timeout: 260 seconds)]
<NixOS_GitHub>
[nixpkgs] volth opened pull request #25332: propagatedUserEnvPackages is obsolete name, these lines have no effect (master...no-effect-propagatedUserEnvPackages) https://git.io/v9C9A
newhoggy has joined #nixos
<jsgrant_>
clever: I'm doing a clean install of NixOS on a newer box & want to run the pre-release on it; Have an init-file from my older box, that is set to 16.09 I plan on/want to use; Wondering if there is a supported way to use nixos-install to install from/with 17.09 instead of the current medium's 17.03.
<clever>
jsgrant_: you can just add whatever channel you want under the name 'nixos' and nix-channel -update
<clever>
jsgrant_: and it should work
<jsgrant_>
Not a huge deal if not, will just comment out most packages for time-being -- but feel like I have to be overlooking.
<clever>
just confirm that there is only one chnanel listed in nix-channel --list
ertes has quit [(Ping timeout: 240 seconds)]
mkoenig has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
newhoggy has quit [(Ping timeout: 240 seconds)]
<jsgrant_>
Ah! Forgot to append '--update'! So used to nixos-rebuild --upgrade doing so automagically; Ty! :^)
samae has quit [(Remote host closed the connection)]
<clever>
2017-03-12 15:43:07< clever> viric: i think the cause was somebody changing how nixpkgs works, and they didnt test pkgs = import <nixpkgs> { platform = pkgs.platforms.foo; }
<Jookia>
clever: Hmm. Is this a bug or a change?
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 5 new commits to master: https://git.io/v9Cdc
<NixOS_GitHub>
nixpkgs/master 1d40717 Jan Malakhovski: firefox: rename default.nix -> common.nix
<NixOS_GitHub>
nixpkgs/master f0f572f Jan Malakhovski: firefox: refactor into firefoxPackages, add more options
<NixOS_GitHub>
nixpkgs/master 2f35ab5 Jan Malakhovski: firefoxPackages: implement privacySupport option
<clever>
Jookia: bug
<Jookia>
Okie
<bennofs>
Jookia: did you use = pkgs.platforms or = pkgs.platform?
eacamero_ has joined #nixos
<Jookia>
bennofs: platforms
Tipping_Fedora has quit [(Ping timeout: 260 seconds)]
newhoggy has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Ping timeout: 252 seconds)]
ma9e has quit [(Remote host closed the connection)]
eacamero_ has quit [(Ping timeout: 260 seconds)]
eacameron has joined #nixos
<NixOS_GitHub>
[nixpkgs] volth opened pull request #25334: xfce4-dockbarx-plugin: init at 0.5 (master...xfce4-dockbarx-plugin) https://git.io/v9Cdo
<NixOS_GitHub>
[nixpkgs] 7c6f434c closed pull request #24733: firefox: refactor, add options; tor-browser: package a version directly usable via nix (master...pkg/firefoxPackages) https://git.io/vSrSs
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 pushed 2 new commits to master: https://git.io/v9CdX
<NixOS_GitHub>
nixpkgs/master 5c7815a Matthew Bauer: impure.nix: add crossSystem as arg
<NixOS_GitHub>
nixpkgs/master f71456a John Ericson: Merge pull request #25301 from matthewbauer/impure-cross...
eacameron has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 240 seconds)]
thineye has joined #nixos
revtintin has quit [(Ping timeout: 255 seconds)]
newhoggy has joined #nixos
<mudri>
Is there a way to get a perl interpreter with libraries in its search path?
<mudri>
The perl documentation only seems to cover compiling perl packages, as opposed to interpreting (AFAICT).
<clever>
error: attribute ‘platforms’ missing, at (string):1:49
<clever>
Jookia: with a ~1.5 year old copy of nixpkgs, it has a different error
<Jookia>
Yesh
<johnramsden>
If I'm setting up a package that needs keyring support, does anyone happen to know how I would use kwallet as a keyring, instead, or in addition to gnome keyring? I couldn't find much information on it.
k2s has joined #nixos
k2s has quit [(Client Quit)]
newhoggy has joined #nixos
peacememories has joined #nixos
newhoggy has quit [(Ping timeout: 252 seconds)]
newhoggy has joined #nixos
thineye has quit [(Quit: Konversation terminated!)]
thc202 has quit [(Quit: thc202)]
aneeshusa has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 260 seconds)]
bennofs has quit [(Quit: WeeChat 1.7.1)]
chaoflow has quit [(Ping timeout: 240 seconds)]
bennofs1 has joined #nixos
Willi_Butz has quit [(Ping timeout: 240 seconds)]
bara has quit [(Ping timeout: 240 seconds)]
bara has joined #nixos
m0rphism has quit [(Ping timeout: 240 seconds)]
avn_nb has quit [(Ping timeout: 240 seconds)]
Willi_Butz has joined #nixos
kosmikus has quit [(Ping timeout: 240 seconds)]
kosmikus has joined #nixos
pSub has quit [(Ping timeout: 240 seconds)]
newhoggy has joined #nixos
pSub has joined #nixos
avn_nb has joined #nixos
m0rphism has joined #nixos
chaoflow has joined #nixos
taktoa has joined #nixos
takle has quit [(Remote host closed the connection)]
takle has joined #nixos
kampfschlaefer has quit [(Ping timeout: 245 seconds)]
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
newhoggy has quit [(Ping timeout: 260 seconds)]
peacememories has joined #nixos
newhoggy has joined #nixos
takle has quit [(Ping timeout: 240 seconds)]
johnramsden has quit [(Quit: Leaving)]
newhoggy has quit [(Ping timeout: 260 seconds)]
peacememories has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
newhoggy has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/v9Cb9
<NixOS_GitHub>
nixpkgs/master 9031c35 Michael Raskin: julia_05: 0.5.0 -> 0.5.1
justanotheruser has quit [(Remote host closed the connection)]
peacememories has joined #nixos
justanotheruser has joined #nixos
ryanartecona has quit [(Quit: ryanartecona)]
peacememories has quit [(Client Quit)]
newhoggy has quit [(Ping timeout: 240 seconds)]
takle has joined #nixos
brandonz has joined #nixos
takle has quit [(Ping timeout: 260 seconds)]
newhoggy has joined #nixos
<sellout->
I know I’ve run into this before, but I get a pile of “download-from-binary-cache.pl: could not download ‘https://cache.nixos.org/xkm84blbsp3fba7rp45ypmwkl519p8zn.narinfo’ (Curl error 60)” errors, and the nix wants to build everything. I remember it involved setting the cert path before, but `export SSL_CERT_FILE=".../ca-bundle.crt"` isn’t fixing it …
newhoggy has quit [(Ping timeout: 268 seconds)]
newhoggy has joined #nixos
<clever>
sellout-: nixos or another distro?
<sellout->
Ah, sorry – on OS X, actually.
<clever>
no nix-daemon in use?
<sellout->
Yeah, I’m using nix-daemon … or should be. Hold on.
<sellout->
Yeah, it’s running.
<clever>
then you need to set that env var for nix-daemon
<clever>
not nix-build
<sellout->
Oh, of course. I realized it as you typed it. Thanks!
newhoggy has quit [(Ping timeout: 255 seconds)]
<sellout->
clever: Perfect. It works. Thanks again.
newhoggy has joined #nixos
justelex has quit [(Ping timeout: 240 seconds)]
newhoggy has quit [(Ping timeout: 245 seconds)]
markus1189 has joined #nixos
markus1199 has joined #nixos
Welkin has joined #nixos
<Welkin>
does nix work for anyone on mac os sierra?
<Welkin>
I keep getting `Illegal Instruction: 4`
newhoggy has joined #nixos
<Welkin>
even while trying to install packages from the cache
<clever>
Welkin: one sec
<clever>
2017-03-13 17:09:26< LnL> echo 'pre-build-hook = ' >> /etc/nix/nix.conf on your osx machine
<clever>
Welkin: this solved some Illegal Instruction issues i was having a month ago
markus1209 has quit [(Ping timeout: 260 seconds)]
markus1219 has quit [(Ping timeout: 260 seconds)]
<Welkin>
I have to create that file?
<Welkin>
it doesn't exist currently
<clever>
yeah
mudri has quit [(Ping timeout: 260 seconds)]
<Welkin>
now I am getting the same error in a different way
<Welkin>
I wonder if I removed nix properly
<Welkin>
just deleting the ~/.nix-* directories and /nix/ is enough, right?
newhoggy has quit [(Remote host closed the connection)]
<clever>
Welkin: id rename them, so you have the option to undo