ma27 has quit [(Ping timeout: 276 seconds)]
phreedom has quit [(Quit: No Ping reply in 180 seconds.)]
phreedom has joined joined #nixos-dev
Sonarpulse has quit [(Ping timeout: 268 seconds)]
globin has joined joined #nixos-dev
orivej_ has quit [(Ping timeout: 255 seconds)]
mbrgm has quit [(Ping timeout: 240 seconds)]
mbrgm has joined joined #nixos-dev
<disasm> ipv6 works now :)
<disasm> issue stemmed from a misunderstanding of accept_ra
<disasm> so you need to specify it on the interface, and if forwarding is allowed you need to set it to 2, setting to 1, true, etc... won't work
<disasm> some would say this could be a security issue blindly accepting routes, but other option is to statically add it
<disasm> meh, sorry for spamming, thought I was in nixos
yorick has quit [(Ping timeout: 248 seconds)]
<disasm> grahamc: I'd also like to discuss support/security team/other stuff with you at some point. Two main reasons, 1) writing a blog post about the state of NixOS for enterprise usage and 2) trying to convince my boss we should start using it for new PM clients.
<grahamc> Great!
ma27 has joined joined #nixos-dev
ma27 has quit [(Ping timeout: 240 seconds)]
jtojnar has joined joined #nixos-dev
<domenkozar> disasm: we run a $3B cryptocurrency with Nix :)
<domenkozar> (this should fit into 1) and induce 2))
orivej has joined joined #nixos-dev
<domenkozar> LnL: so here are my findings
<domenkozar> on darwin, ghc inserts data prefix regardless if data was even populated
<domenkozar> /nix/store/w8jxjn59azbxdzz5n79bs4ya74x8hxqk-stack-1.5.1/share/ghc-8.0.2/x86_64-osx-ghc-8.0.2/stack-1.5.1
<domenkozar> where not even share exists, because stack ships no data
<domenkozar> there's a few ways to fix this, none of them are trivial
<domenkozar> 1) enable data by default (simplest, but requires another full hackage build)
<domenkozar> 2) use dead_strip by default
<domenkozar> 3) fix ghc to refer to data output only when needed
<domenkozar> really the simplest is 1)
<LnL> I thought dead_strip was enabled a while back
<domenkozar> not by default
<domenkozar> also it doesn't play with profiling
<LnL> right
vcunat has joined joined #nixos-dev
<LnL> I've not really looked at the multiple outputs changes, is that enabled by default or just on a case by case basis
<domenkozar> , enableSeparateDataOutput ? false
<domenkozar> , enableSeparateDocOutput ? doHaddock
<domenkozar> , enableSeparateBinOutput ? isExecutable
<domenkozar> , outputsToInstall ? []
<domenkozar> , enableSeparateLibOutput ? true
<domenkozar> , enableSeparateEtcOutput ? (stdenv.lib.versionOlder "7.7" ghc.version)
<domenkozar> see the outlier at the top :P
<domenkozar> I think setting that to true is our best bet
<domenkozar> hmm actually ghc still points to $out even if data is separate output
<domenkozar> so we might need (1)+(3) solutions
goibhniu has joined joined #nixos-dev
<LnL> having an extra data output that's empty is kind of weird, but not a big deal if that fixes it
<domenkozar> cabal2nix sets it to true if data_files is populated
<domenkozar> so really we ought just to fix ghc not to reference it when not used
<LnL> I'm pretty familiar with the haskell expression, but don't really know much about ghc itself
aminechikhaoui has joined joined #nixos-dev
<domenkozar> this rabbit hole is getting huge :)
<domenkozar> LnL: thanks
orivej has quit [(Ping timeout: 250 seconds)]
orivej has joined joined #nixos-dev
__Sander__ has joined joined #nixos-dev
<niksnut> hm, has anybody experienced the NixOS 17.09 AMI in us-east-2 hanging at boot?
<niksnut> ignore that, it looks like it's a network configuration issue
<gchristensen> :o #1700 merged, thanks Eelco!
<niksnut> is there anything else needed for a 1.11 release?
<gchristensen> not that I know of, will this release include aarch64 via nixos.org/nix/install?
<niksnut> I think so
<gchristensen> cool :)
pie_ has joined joined #nixos-dev
<pie_> holy heck lol do you guys have to deal with crap like this? https://www.sigbus.info/software-compatibility-and-our-own-user-agent-problem.html
<gchristensen> pie_: crap _like_ that sure :) however we probably wouldn't shy away from regenerating autoconf scripts :P
<pie_> hm good point
<pie_> :D
<pie_> i suppose that also implies some level of "what could possibly go wrong?" but its not like packages are autogenerated most of the time
<pie_> i suppose immutability also protects packages from eachoher heh
pie_ has quit [(Ping timeout: 255 seconds)]
pie_ has joined joined #nixos-dev
<domenkozar> shlevy: does hydra decl pr testing get tags?
<niksnut> gchristensen: nix build failure on aarch64: https://hydra.nixos.org/build/65928456
<gchristensen> hmm
<gchristensen> I built it on aarch64 prior to sending the PR ...
<Dezgeg> --with-arch=i686 --with-arch=armv8-a --build=i686-unknown-linux-gnu --host=i686-unknown-linux-gnu
<vcunat> :-)
<Dezgeg> does it depend on some funky 32-bit stdenv?
<vcunat> it seems to get to building nix itself, just the test is broken
<Dezgeg> yeah, tests/setuid.nix depends on pkgs.pkgsi686Linux.stdenv
<gchristensen> :/
<vcunat> ah, I see it in the code :-)
<gchristensen> how did I build it okay :/
<gchristensen> niksnut: I'm sorry
<vcunat> that part of tests/setuid.nix should be conditionalized by (system == "x86_64-linux")
<vcunat> (but I don't have an aarch64 to test in anyway)
<sphalerite> gchristensen: ^ :D
<domenkozar> shlevy: aha, one needs to use issue api to get labels on PRs
<domenkozar> :facepalm:
<gchristensen> vcunat: would you like an aarch64 to test with? :)
<vcunat> over long term it would certainly be nice
<vcunat> but I would do the setup in the evening at earliest
<vcunat> now I'm just restarting jobs on Hydra
<gchristensen> hydra isn't even loading for me
<vcunat> hydra.nixos.org seems fine for me ATM
<sphalerite> +1 wfm
* gchristensen shrugs
<sphalerite> are they already rolling out the net-neutrality repeal?
<sphalerite> <_<
<gchristensen> ouch
<Dezgeg> sounds like pkgsi686Linux should be made into an eval error on non-x86
<gchristensen> I can't access hydra. is the setuid test failing on master?
orivej has quit [(Ping timeout: 240 seconds)]
<vcunat> gchristensen: yes, it seems the same on master for aarch64
<gchristensen> ok
<gchristensen> vcunat: can you send me a hashed password? install mkpasswd package and run mkpasswd -m sha-512.
<gchristensen> ok I think I fixed the test, PR: https://github.com/NixOS/nix/pull/1729
<vcunat> "fixed" == removed :-)
<gchristensen> :$
<vcunat> Right, IIRC we don't have good VM-test support for aarch64-linux yet.
pie_ has quit [(Ping timeout: 260 seconds)]
<gchristensen> coming soon
orivej has joined joined #nixos-dev
pie_ has joined joined #nixos-dev
ma27 has joined joined #nixos-dev
<vcunat> 17.09 bumped! Finally I managed it.
<gchristensen> thanks!
<vcunat> unstable coming soon, I hope.
orivej has quit [(Ping timeout: 255 seconds)]
orivej has joined joined #nixos-dev
orivej has quit [(Ping timeout: 240 seconds)]
<clever> Dezgeg: that sounds like a good idea, ive seen the skype package back-fire on darwin, because it grabbed the 32bit linux silently
orivej has joined joined #nixos-dev
<niksnut> gchristensen: binaryTarball.aarch64-linux eval fails: https://hydra.nixos.org/jobset/nix/maintenance#tabs-errors
<gchristensen> "<vcunat> it might be nice if nix supported the substitute-only mode for untrusted users..." <- so untrusted users could be use remote builders
<gchristensen> niksnut: :(
<gchristensen> I should have had the foresight to build every attr in release.nix
<vcunat> My wishlist is unbounded.
<gchristensen> I can't send a patch until after work (8 hours) can someone amend my PR (https://github.com/NixOS/nix/pull/1729) to either: (1) disable shellcheck (https://github.com/NixOS/nix/blob/1.11-maintenance/release.nix#L126-L140) not run on aarch64, or (2) disable shellcheck altogether?
<gchristensen> I could, for example, run a nix-build on all the release.nix attrs after that is updated
<niksnut> I'll disable it
<gchristensen> OK
<vcunat> gchristensen: you know you can fake the authorship/commit time easily?
<gchristensen> lol I do
<gchristensen> I'd rather not do that, because if I get caught it throws all my contributions in to question
<vcunat> oh, right
orivej has quit [(Remote host closed the connection)]
__Sander__ has quit [(Quit: Konversation terminated!)]
orivej has joined joined #nixos-dev
pie_ has quit [(Ping timeout: 240 seconds)]
pie_ has joined joined #nixos-dev
jtojnar has quit [(Read error: Connection reset by peer)]
jtojnar has joined joined #nixos-dev
<niksnut> gchristensen: btw, VM tests on packet-t2-4 seem to fail a lot (always?), e.g. https://hydra.nixos.org/build/65929070
<niksnut> maybe the load is too high
<gchristensen> load avg: 37.85, 41.44, 42.54
<cransom> might be running a bit hot.
<gchristensen> 24 physical cores w/ HT... yeah, 42 is pretty hot
<vcunat> make -l not working correctly?
<gchristensen> not sure
<gchristensen> I can't look at hydra b/c it won't load for me
<gchristensen> but I wonder how many jobs it is configured to run
<vcunat> Some -l bugs were fixed in https://github.com/NixOS/nixpkgs/pull/32271
<vcunat> That's in master now, but not e.g. in 17.09.
<vcunat> You're still blocked from Hydra.nixos.org? Strange.
<gchristensen> vcunat: would you mind if I applied my diff to #32365 and merged tonight?
<vcunat> gchristensen: I think it's a good idea.
<gchristensen> ok, I'll do that unless you beat me to it :)
globin has quit [(Ping timeout: 260 seconds)]
globin has joined joined #nixos-dev
<vcunat> oh, I didn't realize... with 24 cores + HT we surely set --cores 48, so load 42 is a good value.
<vcunat> Well, 42 is a good value for anything... had that room number on the last NixCon, too.
<gchristensen> lol nice
<globin> might want to turn down the cores a bit nonetheless
<vcunat> Why, if it can run 48 threads at once?
Sonarpulse has joined joined #nixos-dev
<gchristensen> well because 24 jobs will each use 48 cores
<vcunat> After we split -j from -l, we can tune this better.
<gchristensen> and until then it is causing lots of VM tests to fail :)
<vcunat> I sometimes get a problem when starting to build on an idle machine with packages that are well-paralelized.
<vcunat> They all fire many threads long before (1-minute) load jumps up.
<Sonarpulse> globin fpletz: https://hydra.mayflower.de/jobset/nixos/pr-29396 the mass rebuild I have in progress from last night is no good can be aborted
orivej has quit [(Ping timeout: 240 seconds)]
orivej has joined joined #nixos-dev
ma27 has quit [(Ping timeout: 246 seconds)]
ma27 has joined joined #nixos-dev
vcunat has quit [(Ping timeout: 250 seconds)]
ma27 has quit [(Ping timeout: 255 seconds)]
ma27 has joined joined #nixos-dev
<clever> https://github.com/NixOS/nixpkgs/issues/30775 has reared its head again, and caused some pretty serious breakage
<clever> i was helping a user lastnight, where the entire desktop manager was broken, because of old QT libs in ~/.nix-profile/
<clever> which left them completely unable to login graphically
goibhniu has quit [(Ping timeout: 255 seconds)]
<domenkozar> peti: I'm almost done preparing multiple-outputs PR
<domenkozar> peti: do you think we could package up hscolor without generic builder so that changing generic builder wouln't always result into 2 recompilations of ghc?
<Sonarpulse> clever: I have a friend that is always hitting catestrophic kde bugs
<Sonarpulse> thanks!
<domenkozar> my feedback loop is ~4h :/
<Sonarpulse> domenkozar: sorry my PR wasn't done for this :/
<domenkozar> if we had prebuilt hscolor, or at least separate derivation, then turnaround would be trivial
<Sonarpulse> domenkozar: you could use a prebuilt hscolor (harcodeded out path) during development at least?
<clever> domenkozar: i have a gist that deals with this, one minute
<clever> that uses the generic-builder.nix from oldpkgs to build the ghc that line 9 uses
<clever> so the ghc itself never depends on the generic-builder.nix your editing
<domenkozar> nice, this is handy
<domenkozar> although solving it properly doesn't really give up much
<domenkozar> clever: thanks
<clever> i wrote that when i was doing the split-output changes
<domenkozar> I'll open an issue to gather feedback
<copumpkin> I've been seeing a lot of this recently:
<copumpkin> line 583: /run/user/0/nix-build-blahblah.drv-0/env-vars: Permission denied
<copumpkin> anyone know what causes that?
<clever> copumpkin: what user owns that file, and what user did nix-shell run as?
<gchristensen> I saw that recently but I chalked it up to being on a system with a too-old-of-a-kernel running too-new-of-a-glibc
<copumpkin> it's something to do with being in nix-shell as root I think
<copumpkin> I exit out of nix-shell and it gives me a sensible output
<srhb> What's better when a backport to stable that has skipped an earlier commit to master in-between? Merge the difference or cherry pick the older commit as well?
<srhb> Here's an example of bringing in an older commit as well, to better explain: https://github.com/NixOS/nixpkgs/pull/32610
<gchristensen> cherry-pick both
<srhb> OK, thanks!
pie_ has quit [(Ping timeout: 240 seconds)]
<copumpkin> argh, the chromium browser derivation really doesn't make it easy to override the version you want
<copumpkin> it calls update.nix, the infamous md5 collision file, which assumes the versions you want are in the same folder as it
* copumpkin pouts
<domenkozar> !m aszlig
<domenkozar> dammit
<copumpkin> opposite of !m
<copumpkin> :P
<copumpkin> It's proving to be very hard to get a different version without changing nixpkgs
<domenkozar> hence the negation
<copumpkin> :)
<copumpkin> !w ?
<copumpkin> oh :)
<domenkozar> !:q
<domenkozar> :!q
<domenkozar> I forgot my vim muscle memory
<{^_^}> You're doing good work, aszlig!
<domenkozar> lol
<gchristensen> is there a function to interpolate an element between every element in a list? ie: `interpolate "foo" [ "bar" "baz" ] == [ "bar" "foo" "baz" ]`
laggy_wifi_ has quit [(Read error: Connection reset by peer)]
laggy_wifi has joined joined #nixos-dev
<copumpkin> do we have libatk-bridge somewhere in nixpkgs?
<domenkozar> gchristensen: yeah there is one
<domenkozar> but I think it's only for strings
<gchristensen> oh?
<gchristensen> oh intersperse
<gchristensen> that is a list.nix function not a strings.nix function! :)
<domenkozar> aha :)
<gchristensen> (but it is in strings.nix, despite being purely list-based)
<jtojnar> copumpkin: libatk-bridge is in at_spi2_atk
<copumpkin> thanks
pie_ has joined joined #nixos-dev
mbrgm has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
<copumpkin> aszlig: have you considered updating the chromium updater to use sha1 collisions? wouldn't want to be behind the times
<clever> :D
<LnL> lol
<gchristensen> you make good points copumpkin
<gchristensen> cryptographically secure hash collisions
<copumpkin> thank you
<copumpkin> I'm always concerned about the security of our updaters
<copumpkin> md5 is known to be insecure
<copumpkin> whereas sha1 is far more expensive to attack
<copumpkin> so we should ergo use sha1
<gchristensen> why do we do that again? why trick nix in to being impure in an updater
<copumpkin> to leak one bit of information
<gchristensen> yeah but why
<copumpkin> I think
<copumpkin> because aszlig didn't want to write a python script?
<copumpkin> to check if a version is available
<copumpkin> (and it's impolite to do impure things from a regular derivation)
<gchristensen> Isee
<domenkozar> "it's easier to break cryptography than write python"
<domenkozar> AMA
<copumpkin> I assume he didn't want to write a python script
<copumpkin> maybe he had other reasons
<gchristensen> I semi-frequently use nix-shell's with shellHooks to update things
<LnL> yeah, I also do that in a few places
jtojnar has quit [(Quit: jtojnar)]
jtojnar has joined joined #nixos-dev
jtojnar has quit [(Remote host closed the connection)]
jtojnar has joined joined #nixos-dev
capisce_ has joined joined #nixos-dev
rsa has quit [(Ping timeout: 248 seconds)]
moredread[m] has quit [(Ping timeout: 248 seconds)]
florianjacob has quit [(Ping timeout: 248 seconds)]
jtojnar has quit [(Ping timeout: 248 seconds)]
_rvl has quit [(Ping timeout: 248 seconds)]
layus has quit [(Ping timeout: 248 seconds)]
capisce has quit [(Ping timeout: 248 seconds)]
jtojnar has joined joined #nixos-dev
_rvl has joined joined #nixos-dev
layus has joined joined #nixos-dev
florianjacob has joined joined #nixos-dev
moredread[m] has joined joined #nixos-dev
rsa has joined joined #nixos-dev
orivej has quit [(Ping timeout: 268 seconds)]
kgz has joined joined #nixos-dev
ma27 has quit [(Ping timeout: 255 seconds)]
orivej has joined joined #nixos-dev
<gchristensen> can anyone sanity-check this PR? https://github.com/NixOS/nixpkgs/pull/32365 I'm about to merge, in about 1hr