gchristensen changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html | 18.03 release managers: fpletz and vcunat | https://logs.nix.samueldr.com/nixos-dev
Sonarpulse has quit [Ping timeout: 256 seconds]
mbrgm has quit [Ping timeout: 252 seconds]
mbrgm has joined #nixos-dev
aminechikhaoui has quit [Ping timeout: 260 seconds]
aminechikhaoui has joined #nixos-dev
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 252 seconds]
lassulus_ is now known as lassulus
drakonis has quit [Remote host closed the connection]
lassulus has quit [Quit: WeeChat 2.0]
lassulus has joined #nixos-dev
octe has quit [Ping timeout: 256 seconds]
jtojnar_ has joined #nixos-dev
jtojnar has quit [Read error: Connection reset by peer]
jtojnar_ is now known as jtojnar
lassulus has quit [Quit: WeeChat 2.0]
lassulus has joined #nixos-dev
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-dev
lassulus has quit [Quit: WeeChat 2.0]
lassulus has joined #nixos-dev
MichaelRaskin has quit [Quit: MichaelRaskin]
jtojnar has quit [Quit: jtojnar]
__Sander__ has joined #nixos-dev
jtojnar has joined #nixos-dev
octe has joined #nixos-dev
<andi-> pstn: as the comments says there is no technical reason not to do it. If it is an important (stable) update then propose a backport (with a PR) and someone can test/merge/comment/…
<srhb> pstn: If you can test it on 18.03 I'll be happy to backport it.
<pstn> srhb: I'm writing you this using the manually backported version and florianjacob from the PR also doesn't seem to have any issues.
<srhb> pstn: Alright :)
<florianjacob> srhb: I was only testing through `services.matrix-synapse.package = unstable.matrix-synapse;`, on top of 18.03 though, but philipp also tested a real backport that should rule out dependencies on other updated python packages or the like.
<srhb> I will open a backport PR and ping you to document this, and then we can merge.
<florianjacob> s/but philipp/but if philipp /
Guest55520 has quit [Quit: WeeChat 1.9.1]
garbas has joined #nixos-dev
<pstn> What about my butt? I wrote down what I did in the PR.
<Mic92> clever: what would bring your own implementation to the table in contrast to a nginx configured as chaching server?
<sphalerite> Mic92: "can also multiplex to several upstream caches" can nginx do that?
<Mic92> sphalerite: you mean load balance?
<sphalerite> no, I mean try each of them until one of them doesn't 404
<sphalerite> so to provide a union of what each cache provides
<Mic92> there is a `try_files` directive for that
<sphalerite> ah, neat
Guest27988 is now known as ocharles
<copumpkin> niksnut: you see my updates to the "claims to be content-addressed but isn't" bug? it's still in 2.0.2, but harder to get
<niksnut> yes
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
aminechikhaoui has quit [Ping timeout: 245 seconds]
aminechikhaoui has joined #nixos-dev
Lisanna has joined #nixos-dev
Jackneill has quit [Quit: Leaving]
<gchristensen> I need some help, I'm seeing some extremely bizarre behavior
<LnL> don't keep us in suspense :)
<gchristensen> writing up a sample
<gchristensen> my Hydra's channel is producing nix expressions where, when I use them, don't actually cause paths to be fetched
<gchristensen> but I can manually `nix-store -r <thepath>` and it is fetched properly
<LnL> that looks like a fake drv
<LnL> > { type = "derivation"; }
<{^_^}> attribute 'drvPath' missing, at (string):2:56
<LnL> err, well if you eval that in the repl you get «derivation ???»
<gchristensen> indeed! Hydra seems to produce fake derivations in its channel
<LnL> builtins.toPath { type = "derivation"; outPath = "/fake"; } #=> "/fake"
Mic92 has quit [Quit: WeeChat 2.1]
<LnL> yeah, so those evaluate to paths but don't have any string context
<gchristensen> right
<gchristensen> one moment
Mic92 has joined #nixos-dev
<gchristensen> LnL: https://gist.github.com/grahamc/907c5fd2dd6d18730128eb36d2bc1319#file-gistfile3-txt this is what is produced by the hydra
<gchristensen> what does your nixexprs look like?
<LnL> mine?
<copumpkin> niksnut: am I doing something wrong in that content-addressable thing, or something that isn't "wrong" but probably should be? just trying to figure out if I can work around it somehow
<gchristensen> LnL: yeah, does your hydra emit similar Nix?
orivej has joined #nixos-dev
<LnL> probably, I don't use the hydra channels
<gchristensen> oh :)
<LnL> don't even remember where to find it :p
<LnL> this is the /latest thing right?
<gchristensen> it'd be at http://foobar.com/channel/latest
<LnL> 'Nix channels are not supported by this Hydra server.'
Sonarpulse has joined #nixos-dev
<gchristensen> this is maddening
<gchristensen> cracked it!
<gchristensen> niksnut, domenkozar: can someone merge this? https://github.com/NixOS/hydra/pull/141
<LnL> wrong url?
<gchristensen> no, that is the right URL
__Sander__ has quit [Quit: Konversation terminated!]
<LnL> oh closed, not merged
<gchristensen> to make it easier, niksnut, domenkozar, can someone merge this? https://github.com/NixOS/hydra/pull/562
aminechikhaoui has quit [Ping timeout: 260 seconds]
aminechikhaoui has joined #nixos-dev
<niksnut> copumpkin: did I understand that this occurs when using a binary cache generated by Nix < 2.0.2 ?
<niksnut> so containing bad CA values
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
<copumpkin> yeah exactly
<copumpkin> I've been running this binary cache for a while
<copumpkin> if the best solution is to go strip CA values out of my existing .narinfo files, I can do that
<copumpkin> but am not sure how to detect which ones are bad, and if it'll break if earlier versions of Nix connect to the cache
<niksnut> yeah, might be best to strip them
<niksnut> I think the bad ones are all that have a "CA" and non-empty References
<copumpkin> ah okay
<copumpkin> and all versions up until 2.0.2 will produce bad CA values in my .narinfo, so I need to upgrade my nix on the binary cache populator
<copumpkin> in an unrelated note, it seems like wrapProgram produces a shebang of #!/usr/bin/env sh which then fails inside sandboxed builds
<copumpkin> but that seems like it would be breaking everywhere so I'm wondering what I'm doing wrong
jtojnar has quit [Quit: jtojnar]
<copumpkin> oh wait, I'm wrong, sorry :)
<LnL> fixupPhase?
<gchristensen> niksnut: is there a reason to not merge https://github.com/NixOS/hydra/pull/562 / the 5yr old PR its based off?
Lisanna has quit [Quit: Lisanna]
<niksnut> no, looks good to me
<gchristensen> yay! thank you!
globin has quit [Quit: o/]
Lisanna has joined #nixos-dev
globin has joined #nixos-dev
orivej has quit [Ping timeout: 265 seconds]
<Sonarpulse> https://github.com/NixOS/nixpkgs/pull/41081 this is a mass rebuild right?
<Sonarpulse> I gather the linux rebuilds are due to some linux kernel thing
<LnL> Sonarpulse: the --extra-framework-dirs error is back since ~4h
<gchristensen> sure wish we could use "not bash" in the installer
<clever> gchristensen: one min
<gchristensen> uh oh
<Sonarpulse> LnL: huh?
<Sonarpulse> ugh
<Sonarpulse> LnL: so I remerged *after* hydra showed all good stufff
<clever> gchristensen: this patches a mach-o binary's dylibs paths so it can run without nix
<clever> gchristensen: so you could download a pre-compiled util that does the bulk of the workload
<gchristensen> oohh
<clever> gchristensen: this one is written in haskell, and does the actual nix install (with your bash script)
<LnL> yeah...
<LnL> Sonarpulse: haskell-updates doesn't test darwin
<Sonarpulse> LnL: :(
<Sonarpulse> LnL: cabal configure --extra-framework-dirs=asdf just worked for me on darwin
<Sonarpulse> 2.0.0.1
<Sonarpulse> LnL: is it cabal-install only?
<LnL> might just be hscolour
<LnL> my instance only builds a small subset
<Sonarpulse> LnL: ./Setup configure --extra-framework-dirs=
<Sonarpulse> for cabal-install 2.0.0.1
<Sonarpulse> which may or may not be older cabal
<Sonarpulse> also parses
FRidh has joined #nixos-dev
<Sonarpulse> LnL: looking
<Sonarpulse> LnL: doubt it
<LnL> doesn't that mean it's built with the binary bootstrap ghc instead of the final one?
<Sonarpulse> LnL: not buildHaskellPackages
<Sonarpulse> that's only different than cross
<Sonarpulse> ghc.bootPkgs.ghc != buildHaskellPackages.ghc in general
<Sonarpulse> it's an unfortunate source of confusion
<Sonarpulse> LnL: have attribute for me to try?
<LnL> the hscolour build of haskellPackages.cabal-install
<Sonarpulse> ok
<Sonarpulse> LnL: so weird this is darwin only!
<Sonarpulse> I have fix I suppose
<Sonarpulse> but let me test moe
<Sonarpulse> *more
<LnL> I still wonder how frameworks work with cross, can you link stuff if you don't have the libraries on the hostPlatform?
<copumpkin> with those library stubs
<copumpkin> of course, if your build then wants to run those programs, you're hosed
<LnL> sure, but that's always a problem with cross
<LnL> I mean result/Library/Frameworks/IOKit.framework/IOKit -> /System/Library/Frameworks/IOKit.framework/Versions/A/IOKit
<copumpkin> yup
<LnL> all you get on another system is the headers and a broken symlink
<copumpkin> oh, yeah, that would need to work differently with .tbd files
<copumpkin> which would work on both platforms, but not be a broken symlink on other cross platforms
<LnL> right so, currently things that depend on (impure) frameworks won't work
<Sonarpulse> tbd is fine for cross
<Sonarpulse> broken symlink is fine
<Sonarpulse> you can do pure tbd too
<Sonarpulse> I'd love if we did that for all libraries
<LnL> what, the symlink is fine?
<Sonarpulse> so we only rebuilt on abi change
<Sonarpulse> with intensional store
<Sonarpulse> yeah build system never needs to deref symlink
<Sonarpulse> so w/e
<LnL> ld doesn't need to open the file?
<Sonarpulse> nop!
<Sonarpulse> dynamic linking works just fine
<LnL> oh, I expected that to be neccecary
contrapumpkin has quit [Ping timeout: 245 seconds]
FRidh has quit [Quit: Konversation terminated!]
Lisanna has quit [Remote host closed the connection]
<LnL> Sonarpulse: testing
<Sonarpulse> LnL: not i just pushed again
<LnL> lol
<Sonarpulse> LnL: *note
<Sonarpulse> fiddled with the condition
<LnL> hmm why !buildPlatform.isDarwin and not targetPlatform.isDarwin
<Sonarpulse> LnL: I don't really know why the issue
<Sonarpulse> but yeah sure try making it hostPlatform
<Sonarpulse> LnL: oh buildPlatform because we are building setup.hs
<Sonarpulse> err awit
<Sonarpulse> haha nope
<Sonarpulse> err wait yes
<LnL> oh
<Sonarpulse> nativeGhc is what gives us the cabal library
<Sonarpulse> the cabal library is multi target
<copumpkin> Sonarpulse: wait, I'm pretty sure ld does need to open the file
<copumpkin> because library id is stored within it
<Sonarpulse> copumpkin: ios tbd works
<Sonarpulse> no ios libs shipped
<copumpkin> like, 1) tell ld you want to link /lib/foo.dylib 2) ld looks in /lib/foo.dylib, says, "hey foo, what's your name" 3) foo says my name is "bar.dylib" 4) ld says, okay, I'll include "bar.dylib" in the binary I generate
<copumpkin> oh yeah, I'm in response to you saying the broken symlink is fine
<copumpkin> I don't see how it could be fine given the semantics for library id
<copumpkin> I was proposing .tbd files so I know those work :)
<Sonarpulse> copumpkin: oh i guess I didn't mean borken symlink but dangling tbd reference
<Sonarpulse> +1
<Sonarpulse> dyblib filepath to nowhere
<copumpkin> oh yeah, that's fine and what .tbd files were made for
<copumpkin> but what we have right now are symlinks with broken targets
<copumpkin> libtapi is now on opensource.apple.com too
<Sonarpulse> copumpkin: oh btw
<copumpkin> so this is all doable fairly purely
<Sonarpulse> i wrote to mailing list
<Sonarpulse> about libtapi
<Sonarpulse> it was tried to be upstreamed before
<copumpkin> oh?
<Sonarpulse> then there was a pause
<Sonarpulse> then i tried to get going again
<Sonarpulse> seem to get some approval
<Sonarpulse> but no time to do the work yet
<copumpkin> oh, on LLVM?
<Sonarpulse> we're overdue getting our stuff on latest nixpkgs haha
<Sonarpulse> yup!
<Sonarpulse> the author of TAPI made some more changes in his own github repo beyond the open source release
<copumpkin> oh okay
<copumpkin> cool
<Sonarpulse> if you would wanna split the work on that, would be very nice :)
ixxie has joined #nixos-dev
MichaelRaskin has joined #nixos-dev
drakonis has joined #nixos-dev
drakonis has quit [Remote host closed the connection]
drakonis has joined #nixos-dev
<gchristensen> it feels so good landing patches to makestuff easier
<drakonis> do tell
<gchristensen> just fixing little pain points
<drakonis> nice.
ixxie has quit [Ping timeout: 245 seconds]
pie_ has quit [Remote host closed the connection]
drakonis has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
drakonis has joined #nixos-dev
Sonarpulse has quit [Ping timeout: 265 seconds]
globin has quit [Quit: o/]
globin has joined #nixos-dev