<bgamari> right
<Sonarpulse> which is the error we used to get in build != host == target gcc
<bgamari> we probably need to override a few more autoconf checks
<Sonarpulse> for the arm ones
<bgamari> yes
<Sonarpulse> bgamari: well it worked without any overrides before
* bgamari is quite skeptical of this
<bgamari> I don't see how that is possible
<Sonarpulse> well, it at least built
<Sonarpulse> Dezgeg would know if the thing that built actually worked :D
<bgamari> it almost certainly didn't cross-compile
<bgamari> gcc has a number of autoconf checks that require running host code
<Sonarpulse> yeah there is a thing that that skips those tests
<Sonarpulse> for autoconf in general, I thought
<bgamari> I don't believe so
<bgamari> It needs to know what the result would be
<Sonarpulse> woo!
<Sonarpulse> ok I just built gcc build != host == target
<Sonarpulse> on cross elegant
<Sonarpulse> without them
<Sonarpulse> :D
* bgamari is confused
<Sonarpulse> did you fix gawk?
* Sonarpulse diffs
<bgamari> yes
<bgamari> it looks like I had to define AR
<bgamari> Sonarpulse, ^
<Sonarpulse> bgamari: hmm weird weird
<Sonarpulse> oh makeFlags
<Sonarpulse> cool
<bgamari> righto
* bgamari also managed to get boost cross-compiler
<Sonarpulse> this is going to mass-rebuild me, hah
<bgamari> after much wrangling
<bgamari> hmm
<bgamari> unrelated to cross-compilation
<bgamari> I am seeing a fetchgit derivation fail with "No user exists for uid 1000"
<bgamari> which is odd since 1000 is my uid
<bgamari> which most certainly exists
<Sonarpulse> bgamari: user namespace perhaps?
<Sonarpulse> nix sandboxing
<Sonarpulse> hmm now just need diffutils to pass
<bgamari> Sonarpulse, I have a patch for that too
<Sonarpulse> oh
<Sonarpulse> didn't see
* bgamari pushes
<Sonarpulse> oh page 2
<Sonarpulse> haha
<bgamari> 32d1d17d378fb11b4467126ead90ba1623d9b3c5
<bgamari> perhaps that too
* bgamari needs to squash
<Sonarpulse> bgamari: oh ok hah just got that
<Sonarpulse> bgamari: what's the other diffutils commit?
<Sonarpulse> didn't see
<Sonarpulse> (or can wait for you to squash)
<bgamari> it looks like there's actually only one
* bgamari just assumed we were referring to two different commits
<Sonarpulse> huh
<Sonarpulse> will look more closely at error then
<bgamari> what is the error?
<Sonarpulse> cp -d /nix/store/g8iz92324r4bjdisqk9qaw5f625jp5f5-stdenv-bootstrap-tools-cross-aarch64-unknown-linux-gnu/lib
<Sonarpulse> cp: missing destination file operand after '/nix/store/g8iz92324r4bjdisqk9qaw5f625jp5f5-stdenv-bootstrap-tools-cross-aarch64-unknown-linux-gnu/lib'
<Sonarpulse> weird one
<bgamari> mm
<bgamari> that is an odd one
<bgamari> I've not seen that
<bgamari> huh, even with `--option build-use-sandbox false` I still get this `No user exists` error
<clever> if it has root, it will always try to drop root
<clever> sandboxing is a seperate thing
<bgamari> I'm running `nix build` as `ben`
<clever> does ben have +w to /nix/store/ ?
<bgamari> yep
<bgamari> ben owns /nix/store
<clever> then it should just directly build things as ben
* bgamari is using nixUnstable for the record
<LnL> without the daemon?
* bgamari is on debian, so yes
<bgamari> I haven't installed the daemon
<bgamari> afai
<bgamari> k
<LnL> try with --option store local
<bgamari> same issue
<Sonarpulse> oh the error isn't diffults, yay!
<Sonarpulse> libelf was just being statically linked, odd
<bgamari> hmm, --debug doesn't really offer much of use
<Sonarpulse> bgamari: "configure: warning: shared libraries not supported for aarch64-unknown-linux-gnu"
<Sonarpulse> seen anything like that?
<bgamari> it's possible
<bgamari> I m may have overlooked it
<bgamari> ahhh
<bgamari> [pid 13121] openat(AT_FDCWD, "/nix/store/mjx71lmnlf4psm9942djjcd8b56hyk8b-glibc-2.26-75/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
<bgamari> how could this have even happened
<Sonarpulse> weird
orivej has quit [(Ping timeout: 248 seconds)]
Sonarpulse has quit [(Ping timeout: 255 seconds)]
<clever> bgamari: nixos doesnt have an ld.so.cache
<bgamari> clever, figured it out eventually
mbrgm has quit [(Ping timeout: 240 seconds)]
mbrgm has joined joined #nixos-dev
orivej has joined joined #nixos-dev
<bgamari> SonarPulse: Why does the bintools wrapper not expose ar, ranlib, and strip?
orivej has quit [(Ping timeout: 240 seconds)]
orivej has joined joined #nixos-dev
ma27 has joined joined #nixos-dev
vcunat has joined joined #nixos-dev
JosW has joined joined #nixos-dev
FRidh has joined joined #nixos-dev
_rvl_ has joined joined #nixos-dev
_rvl has quit [(Ping timeout: 252 seconds)]
<domenkozar> LnL: found it
<domenkozar> $ grep -R /nix/store/3l7i7yjq4chb0iach10vyvnnxndga76v-hevm-0.8.5 /nix/store/rwr25czyplg3xi4pvgpzrkl880ymi878-hevm-0.8.5-bin
<domenkozar> Binary file /nix/store/rwr25czyplg3xi4pvgpzrkl880ymi878-hevm-0.8.5-bin/bin/.hevm-wrapped matches
<domenkozar> plot thickens :)
<domenkozar> ]AA]A^A_Ðf.f.@HHtH1,f.1,HH/nix/store/3l7i7yjq4chb0iach10vyvnnxndga76v-hevm-0.8.5/share/ghc-8.0.2/x86_64-linux-ghc-8.0.2/hevm-0.8.5hevm_datadirtimeouterrormainMainhevm-cli/hevm-cli.hsinternal error; no EVM result..t.sol.json, Files found: Specify one using `--json-file'
<domenkozar> huh, this should end up in data output
clever has quit [(Ping timeout: 248 seconds)]
clever has joined joined #nixos-dev
clever has quit [(Ping timeout: 255 seconds)]
clever has joined joined #nixos-dev
clever has quit [(Quit: Reconnecting)]
clever has joined joined #nixos-dev
JosW has quit [(Quit: Konversation terminated!)]
vcunat has quit [(Ping timeout: 246 seconds)]
<bgamari> How can I get the absolute path to the in-scope target ar from a setup hook?
<gchristensen> $(which ar) ?
<bgamari> well, that would require that `which` sis in scope, no?
<orivej> command -v ar
<bgamari> hmm
<bgamari> alright, sounds plausible, let's see how this goes
vcunat has joined joined #nixos-dev
clever has quit [(Changing host)]
clever has joined joined #nixos-dev
<gchristensen> we're getting big enough to get "spam" security advisories
<bgamari> will things blow up terribly if I run multiple nix-build processes simultaneously?
<gchristensen> no
<gchristensen> definitely not
<bgamari> lovely
<bgamari> will they avoid building packages in their transitive closures twice?
<gchristensen> yes
<bgamari> cool
<gchristensen> you can even kick off a nix-build, then once it starts building change what the source code in the dir it is building without impacting the current nix-build
<gchristensen> ikwildrpepper, niksnut: either of you to validate some credentials are bogus?
<gchristensen> aszlig: ping
<vcunat> gchristensen: what are "spam security advisories"
<vcunat> CVE numbers that claim incorrect things?
<gchristensen> people who search for "AKIA" and send reports wanting to be in your project's 'security reporter hall of fame'
<gchristensen> (akia being the prefix to aws access keys)
<bgamari> is there any way to ensure that a builder is run as root?
<bgamari> I understand that this is a bit hairy given that it might trash the store
<gchristensen> these questions are more appropriate for #nixos
<bgamari> thanks
<gchristensen> (where I'll gladly answer your questions)
<gchristensen> vcunat: 5 days later on check meta
<vcunat> I must say I'm really tired today
<gchristensen> oh, take care of your self! :) there are more important things than nixos
<vcunat> well, today it wouldn't be efficient, regardless of importance
<gchristensen> hrmm it is a bummer eelco hasn't released an updated nix for high sierra
vcunat1 has joined joined #nixos-dev
vcunat has quit [(Ping timeout: 240 seconds)]
vcunat1 has quit [(Client Quit)]
vcunat has joined joined #nixos-dev
<ikwildrpepper> gchristensen: what credentials?
clever has quit [(Ping timeout: 240 seconds)]
clever has joined joined #nixos-dev
clever has quit [(Changing host)]
clever has joined joined #nixos-dev
jtojnar has quit [(Quit: jtojnar)]
jtojnar has joined joined #nixos-dev
<orivej> May someone restart aborted jobs in https://hydra.nixos.org/eval/1417461 ?
<vcunat> orivej: that was me
<vcunat> I chose to prefer jobsets that rebuild with the updated openssl
<vcunat> (with security fixes)
<gchristensen> oops I restarted them :) I just cancelled them again
<vcunat> it's only a thousand jobs, so I didn't know if it's worth cancelling anyway
<laggy_wifi_> in a dynamic library is there any section/byte range in wich a NEEDED depandancy is garenteed to be located in?
<gchristensen> laggy_wifi_: #nixos-dev is not the place to ask about things unrelated to nixos dev.
<laggy_wifi_> ok
Sonarpulse has joined joined #nixos-dev
<bgamari> Sonarpulse, I have cross-compiled gcc
<Sonarpulse> bgamari: nice!
<aszlig> gchristensen: pong
<gchristensen> aszlig: someone emailed me, unhappy that we have google api keys in nixpkgs for chromium. do you know if those API keys are valid / what they are capable of doing / why they're there?
<aszlig> gchristensen: what was the complaint exactly?
<samueldr> AFAIK, it's all pretty cromulent to have them in there
<gchristensen> nice, thank you aszlig
<gchristensen> aszlig: it was a security researcher reporting private key exposure
<aszlig> X-D
<gchristensen> :)
<vcunat> :-D
<gchristensen> whoa I just got a rare emerge from Amine
<gchristensen> rare merge*
<aminechikhaoui> seemed trivial enough to merge :D
<gchristensen> :)
<domenkozar> anyone around that knows about ghcjs?
<Sonarpulse> domenkozar: it's been a while
<Sonarpulse> but might know something
<Sonarpulse> (since I've touched the nix side of ghcjs)
<domenkozar> each binary gets a topdir wrapper, I guess that's needed to touch lib
<domenkozar> actually I think I solved it :)
<domenkozar> Sonarpulse: sorry for using you instead of duck
<domenkozar> rubber duck*
<domenkozar> Sonarpulse: what do you think about https://github.com/NixOS/nixpkgs/issues/29011
<domenkozar> we could eliminate justStaticExecutables
<Sonarpulse> domenkozar: haha no worries
<Sonarpulse> domenkozar: do we do any optimization by default?
<domenkozar> for executables?
<Sonarpulse> yeah
<Sonarpulse> I suspect that would tip the scales on dynamic vs static for lots of haskell executables all using the same libraries
<domenkozar> scales meaning size?
<Sonarpulse> domenkozar: yeah. If we don't do a lot of cross-package inlining, dynamic could well be smaller overall
<Sonarpulse> (but yes very annoying in single exe case)
<Sonarpulse> if we do a lot of cross-package inlining, then screw dynamic
<Sonarpulse> perhaps the answer is we don't, precisely because dynamic
<domenkozar> it's more about the fact that you'd install executables
<domenkozar> and you want the closure to be small
<domenkozar> as soon as something depends on haskell lib, you bring in ghc with 1GB output size
<domenkozar> and dynamic executables are like a few MB, while static are a few 10MB
<domenkozar> at least by my imperic biased observation :)
<Sonarpulse> domenkozar: so base in ghc is just bad, heh
<Sonarpulse> I guess the question is how many haskell exes
<Sonarpulse> before closure size is the same for both
<Sonarpulse> assuming ghc is no in closure either way
<domenkozar> in case of dynamic libs it's much much bigger
<domenkozar> you can't beat dead code elimination with multiple outputs
<clever> it would need to be one output per symbol, lol
<clever> overhead would be murder
<clever> which reminds me, embeded avg-glibc handles dead code elimination by putting every function into its own .o, all packaged into a .a
<clever> then at link time, the linker puts a subset of the .o files
<domenkozar> sounds like a great way to reach max size limit on ext4
FRidh has quit [(Remote host closed the connection)]
<domenkozar> Sonarpulse: btw is sierra hack by default on?
<Sonarpulse> domenkozar: no it is not
<Sonarpulse> domenkozar: it also needs to be merged into 17.09 and master :(
<Sonarpulse> former has PR open
<Sonarpulse> latter I didn't do yet, in hopes someone was going to just get rid of --rpath on linux
<Sonarpulse> which would simplify a few things
<Sonarpulse> but perhaps I should stop waiting or that
<domenkozar> LnL: then I don't understand why you think macOS hack is the reason that happy fails with bin output
<domenkozar> hmmz :)
<domenkozar> wow I got ghcjs to work with multiple outputs
<domenkozar> so it's just macOS left
<domenkozar> clever: damn GC is faster than I can build the whole ghc pipeline on our macs
<domenkozar> :'(
<domenkozar> clever: iirc it's every hour?
<clever> let me double-check
<clever> yeah, hourly, and it stops at 25gig free
<domenkozar> out of 45GB
<domenkozar> :facepalm:
<clever> the mac i checked has 16gig free right now
<Sonarpulse> orivej: I'm clearly very bad at fixing this on the fly... :/
<Sonarpulse> orivej: when I rebased cross-elegant
<orivej> Sonarpulse: is it urgent?
<Sonarpulse> I went back to single getOutputs
<Sonarpulse> it's dead code in the native case, so no
<Sonarpulse> but then on 3rd thought
<Sonarpulse> decided that was probably fine
<domenkozar> sadly macincloud charges 10% per 50GB for additional storage
<domenkozar> 10$*
<domenkozar> quite a steal
<Sonarpulse> orivej: https://hydra.nixos.org/jobset/nixpkgs/cross-trunk see that lack of `or {}` messed things up
<Sonarpulse> ok now it works
<orivej> Sonarpulse: which job at cross-trunk should I look at?
phreedom has quit [(Ping timeout: 240 seconds)]
<Sonarpulse> orivej: nix-instantiate pkgs/top-level/release-cross.nix -A bootstrapTools.aarch64.build
<orivej> Sonarpulse: is it correct that "fetchpatch" is missing from "runPkgs"?
<Sonarpulse> orivej: yes
<orivej> why?
<Sonarpulse> I purposely filtered out all the fetch* functions
<Sonarpulse> I would need to change every use-case to buildPackages.fetch* otherwise
<Sonarpulse> don't want to use cross-compiled cURL!
<Sonarpulse> orivej: so longer term
<Sonarpulse> 1) splicing sucks
<Sonarpulse> 2) package should be able to specify their "default usecase" for people that can't bother to use the right sort of dependency manually
<Sonarpulse> the fetch* filter thing is like an ad-hoc first attempt at 2)
vcunat has quit [(Quit: Leaving.)]
vcunat has joined joined #nixos-dev
<orivej> Sonarpulse: could you fix gnat in staging ( https://github.com/NixOS/nixpkgs/pull/32403#issuecomment-350535317 )?
<Sonarpulse> orivej: just doing that! :)
vcunat has quit [(Quit: Leaving.)]
<Sonarpulse> niksnut: the main cross PR is quite close, so it's all the more important that the bintools-wrapper PR, https://github.com/NixOS/nixpkgs/pull/29396, be merged
<Sonarpulse> I hope you will have time to review that soon
<orivej> Sonarpulse: have you fixed the build of "libstdcxx5" in #29396?
<Sonarpulse> orivej: oh right
<Sonarpulse> orivej: I was going to after the last nix-borg
<Sonarpulse> but that was aborted for some reason
<orivej> except for the build failures, that PR looks good to me
<Sonarpulse> orivej: thanks :)
<Sonarpulse> I want niksnut to sign off on it thought because he reverted it before!
<gchristensen> why did he revert it before?
olejorgenb[m] has left #nixos-dev ["User left"]
<Sonarpulse> gchristensen: because it was a big change
<Sonarpulse> he wanted to personally review
<gchristensen> ah
<gchristensen> ok a slightly scary ofborg deploy, fingers crossed :)
{^_^} has quit [(Remote host closed the connection)]
{^_^} has joined joined #nixos-dev
{^_^} has quit [(Changing host)]
{^_^} has joined joined #nixos-dev