<clever>
so if you somehow had 1tb of swap, and tried to use more then 1tb of haskell heap, it would fail
<clever>
i also recently thought of a potential exploit in that, reading the source, its not capable of allocating more space when that 1tb runs out
<clever>
instead of constantly trying to allocate memory from the kernel and pay the overhead, it just grabs a 1tb chunk, and lets lazy allocation deal with it
<clever>
haskell programs for example, i think use 1tb of virtual memory
<clever>
if your text editor needs a daemon to hide the load times, it has issues
<clever>
kuznero: does it use ghc-pkg?
<clever>
kuznero: what does ghc-mod normally do behind the scenes?
<clever>
kuznero: depends on how well you know haskell i think, i had some trouble getting it to work because the types where confusing at first
<clever>
i just stick to vim
<clever>
ah
<clever>
nix-mode i think it was called
<clever>
kuznero: i'm checking this to see what it does to ghc-mod
<clever>
oh, yeah, last change to that file, 23 hours ago
<clever>
that part looks valid as well
<clever>
does the error mention any other files or line numbers?
<clever>
but i dont see anything abnormal with the latest version on master
<clever>
that is what is activating that file
<clever>
romildo: do you have programs.zsh.syntaxHighlighting.enabled set?
<clever>
evangeline: --check can force a given derivation to rebuild, but the entire goal of nix is to track what can affect the build, and automaticaly rebuild when that has changed
<clever>
evangeline: i believe the only option right now is to manualy read the file, its just text
<clever>
evangeline: it was removed several months ago
<clever>
and gcc has a setup-hook that will add everything in $buildInputs to the -I path
<clever>
because installing things doesnt run the setup-hook
<clever>
but if you do list it, it will just work
<clever>
installing gcc breaks a lot of the features nix adds to help you
<clever>
Filystyn: you need to add gcc to the -p list as well
<clever>
you can paste code into that {} that will call it
<clever>
but that just tells nix-instantiate to eval the {} at the end and create its .drv file
<clever>
evangeline: that isnt inside an attrset or let block, so its not valid
<clever>
Filystyn: what nix-shell args are you using?
<clever>
nix-build builds it for you, acording to directions in a nix expression
<clever>
nix-shell gives you a shell for building
<clever>
Filystyn: your suppossed to do everything inside either nix-shell or nix-build
<clever>
Filystyn: your not supposed to install libraries on nix
<clever>
evangeline: these arguments can be passed in
<clever>
gleber_: is escript in $PATH at that time?
<clever>
gleber_: as long as the file exists during the patchphase, it should work
<clever>
gleber_: yeah, patchSheBangs will fix the problem, once you know which file, oh, its make/emd2exml, so just postPatch = "patchSheBangs make"; i think?
<clever>
gleber_: not really, but you could temporarily delete /usr/bin/env from the host, thats not likely to cause too many problems
<clever>
gleber_: but even with nix-shell --pure, you still have a /usr/bin/env
<clever>
gleber_: nix builds are sandboxes, so they dont have a /usr/bin/env
<clever>
gleber_: what is the error its giving?
<clever>
gleber_: the shell made by nix-shell also sources setup for you, so thats not required
<clever>
gleber_: nix-shell launches a new shell, so the ; at the end wont do waht you want
<clever>
xificurC: yeah, you must always use nix-store --delete to remove things from the store
<clever>
xificurC: try it on a path your cache has that you dont
<clever>
xificurC: nix-store -r /nix/store/foo will work on any path in a binary cache
<clever>
sphalerite: '<nixpkgs>' will resolve to the path of nixpkgs
<clever>
i got an email today about multi-mc failing due to a dep
<clever>
dihuteno: i think you can do services.openssh.extraConfig = lib.mkForce ''.....''; to do that
<clever>
dihuteno: what exactly are you trying to do?
<clever>
dihuteno: thats part of the nixos module, not the package, so it has to go into services.openssh
<clever>
gchristensen: i'm guessing its a haskell thing
<clever>
dihuteno: yeah
<clever>
dihuteno: nixos will merge all of them together
<clever>
dihuteno: you can list extra modules in the imports section, and they have the identical syntax to configuration.nix
<clever>
neoo: when you import nixpkgs like that, it reads the config.nix from $HOME, but you can force a custom config like i just showed
<clever>
neoo: yeah, line 12 is the problem, you want to put this at the end of 6, { config.allowUnfree = true; }
<clever>
oh wait, and he is doing that in a weird way
<clever>
FRidh: he is already doing that, but its using a different config.nix
<clever>
gleber_: is it being ran during buildPhase as well?
<clever>
xificurC: callPackage ./as-table {};
<clever>
gleber_: i would do it in preConfigure or postPatch
<clever>
but you can manualy run the patchSheBangs function on any directory
<clever>
fixup is after install
<clever>
gleber_: its done by the patchSheBangs function, that runs during the fixup phase
<clever>
jophish: if a temproot exists for a pid that does not, id call that a bug, and you should be able to safely rm -rf the temproot
<clever>
nice
<clever>
is process 21274 still alive?
<clever>
ah yeah
<clever>
the --roots really needs to not hide those invisible roots
<clever>
yeah, not sure if that query exists
<clever>
what about its referers?
<clever>
ah, try the referrers hash in the above command then
<clever>
that sounds like sqlite
<clever>
you should see the open syscall being used on a /proc path, then a read() syscall that contains the hash
<clever>
elehack: as long as you can remotely force a reboot, or ensure it reboots upon failures
<clever>
tanonym: not sure why
<clever>
KABA: while LD_LIBRARY_PATH just changes the search path, just be sure to prepend/append, because opengl also needs LD_LIBRARY_PATH to stay intact
<clever>
KABA: LD_PRELOAD can potentialy break some programs that your program may spawn, since it forces things to load
<clever>
tanonym: oh, i see the problem, its 32Bit, not 32bit
<clever>
tanonym: can you gist your configuration.nix file?
<clever>
because it goes to the 2nd oldest, not the previous
<clever>
rollback sets current = current-1, but if you do update, rollback, update, rollback: it will wind up on the wrong generation
<clever>
maybe rollback would work for the 2nd one, but thats got a minor issue when you go forward and back often
<clever>
elehack: yeah, you would sort of need to run nixos-rebuild boot twice, and have the 2nd time undo the change, creating a "dead" previous generation, that you could grub-reboot to
<clever>
elehack: that tool lets you do one-time changes to the default boot entry
<clever>
elehack: you can maybe do something with grub-reboot, but you would need to experiment on a box with a monitor
<clever>
jophish: nix-serve can sort of do that already, but you would have to add every machine as a binary cache
<clever>
arianvp2: neat
<clever>
m0rphism: as long as you have +w on the directory, you can probably delete/replace anything
<clever>
m0rphism: vim will save to a temporary filename, then use rename to atomicly replace the file
<clever>
even if you dont have physical access
<clever>
you can basicaly turn any linux machine into a nixos machine
<clever>
pseudo-sue: this makes dumpcap setuid root, only executable by members of the wireshark group, creates the group, and puts my user in the group
<clever>
Drakonis[m]: ive run into similiar problems on gentoo
<clever>
ive had an ubuntu server fail mid way thru dist-upgrade, and leave a mess everywhere
<clever>
allowing multiple glibc's also allows you to update glibc without the risk of breaking things
<clever>
but nix-env can make it easy to have dozens of apps, spread over dozens of glibc's
<clever>
as long as you install everything in a declarative way, you will only have 2-4 glibc's (32bit, 64bit, possibly over 2 channels like your doing)
<clever>
joehh: nix uses absolute paths for everything, i think i have 8 to 10 glibc's on my desktop
<clever>
yeah, because you imported nixos-unstable, it wants the nixos-unstable version of glibc, and everything else dropbox depends on
<clever>
the example ToxicFrog originaly gave uses the /root/.config/nixpkgs/config.nix to set unfree
<clever>
joehh: what is in /root/.nix-defexpr/channels/
<clever>
joehh: did you --list with or without sudo?
<clever>
thats how things would work if you use nix-channel
<clever>
you can put a url into configuration.nix, but then nix may update it without warning
<clever>
its like #include <stdio.h>
<clever>
you do need it
<clever>
yeah
<clever>
drakonis: [x]
<clever>
in the directory you where in when you ran nix-build
<clever>
it creates a symlink called result by default
<clever>
nope
<clever>
yeah
<clever>
drakonis: does "git remote -v" show your fork?
<clever>
drakonis: when you push your changes to github, that will update it
<clever>
you need to push to a fork you made on github, then use the ui in the browser
<clever>
there is a nix-prefetch-url for when its a simple url
<clever>
drakonis: nix doesnt hash the bare tar in most cases, so you need to use the hash nix-build spits out
<clever>
:)
<clever>
that should help in manually running dhcpcd to get an ip
<clever>
you should be able to find the path to dhcpcd
<clever>
capisce: if you run nix-store -qR /run/current-system | grep dhcp
<clever>
anything in "journalctl -f -u wpa_supplicant" ?
<clever>
is the wpa config in /etc/wpa_supplicant.conf ?
<clever>
capisce: stop it, then run "systemctl start wpa_supplicant"
<clever>
capisce: did you run the wpa service or manualy start wpa_supplicant?
<clever>
capisce: nixos uses dhcpcd, it works pretty much the same way
<clever>
LnL: so i dont think hydra can do everything in a real channel
<clever>
LnL: programs.sqlite is an IO heavy part of the channel that needs external access
<clever>
the nix-channel manpage also shows that, yeah
<clever>
it just needs to be an http directory containing 2 files, binary-cache-url and nixexprs.tar.xz
<clever>
not sure, i havent really used perl since 2005, and even then, i'm not sure if ive ever seen a perl repl, lol
<clever>
and 19, tell it to read the config
<clever>
jophish: i think you need to import nix first
<clever>
jophish: any env variables starting with NIX_ set in the daemon context?
<clever>
i think each worker the daemon spawns will re-read the file, so restarting isnt required
<clever>
jophish: where is the nix.conf file?
<clever>
it shouldnt
<clever>
jophish: ah, what happens if you just remove cache.nixos.org entirely?
<clever>
jophish: i believe its purely on the order, and if you use mkForce, it will bypass the default value, so you can add cache.nixos.org near the end
<clever>
i'm also using a hangouts extension in chrome
<clever>
and people still stick with it, because they think skype is better
<clever>
so M$ bought skype, and then turned it into MSN
<clever>
something i heard years ago, is that nobody liked MSN, but everybody liked skype, because of the history with how good it was
<clever>
lol
<clever>
ah
<clever>
i cant even see the images people embed in chat without having to sign in on a browser for every single image
<clever>
M$ has been slowly ruining linux support
<clever>
ive not tried the new skype yet
<clever>
ah
<clever>
yeah, i cant find it on the old skype 4.3
<clever>
oh wait, thats where they hide it on windows, lol
<clever>
under the call menu is share your screen
<clever>
Nobabs27: skype and hangouts both have screen sharing without control
<clever>
Drakonis[m]: ah yeah, i think i saw you mention it yesterday?
<clever>
Drakonis[m]: make a nix file sort of like this and then run nix-shell against it: with import <nixpkgs>{}; clangStdenv.mkDerivation { name="name"; buildInputs = [ libpcap ]; }
<clever>
Drakonis[m]: there is a clangStdenv as well to handle that
<clever>
deba5e12: strange, it only looked in the libc directory, and nowhere else
<clever>
Drakonis[m]: if your using normal gcc and nix-shell, it just works with things like -lpcap
<clever>
deba5e12: ive seen a java engine do that as well
<clever>
deba5e12: next thing i would do is to run the failing command under "strace -ff -o logfiles -e open ..." and then see what paths its trying to open
<clever>
what is the exact error it gives?
2017-06-01
<clever>
dang
<clever>
try running the lisp repl inside the above nix-shell and see if that helps