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
<dtz> volth tearing it up on the Nix perf/mem improvement PR's! :D
<gchristensen> w00t
<gchristensen> how the heck does staging ever build with the daily big ol changes
<dtz> is volth on IRC under a different name?
<dtz> lol
<dtz> no idea
<dtz> we need building Nix packages to be the "work" done by some blockchain thing that takes off
<dtz> nix build --option builders bitcoin
<dtz> lol
Sonarpulse has quit [Ping timeout: 265 seconds]
phreedom_ has joined #nixos-dev
phreedom has quit [Ping timeout: 250 seconds]
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 260 seconds]
lassulus_ is now known as lassulus
Lisanna has quit [Quit: Lisanna]
phreedom_ has quit [Ping timeout: 250 seconds]
phreedom has joined #nixos-dev
Lisanna has joined #nixos-dev
orivej has joined #nixos-dev
tv has quit [Ping timeout: 256 seconds]
tv has joined #nixos-dev
<pie__> dtz, oh god stop lol
<pie__> had to reread to get it
<pie__> well at least theyd be causing global warming with something useful :D
<andi-> nixcoin, the more derivations you build the more coins you get. Chromium gives the higscore? m(
<clever> andi-: sounds like a great reason to split chromium up into 1 derivation per .o file!
<andi-> incremental builds using nixos?
<andi-> to bad the makefile or whatever probably will be touched/changed an all the others rebuild as well..
<clever> andi-: you would need to parse the makefile, and the #include statements, and compute the entire dependency graph
<clever> andi-: then generate a derivation tree, with a sandbox of .h files for each .o
<clever> then you get incremental builds
<andi-> I know.. was just a bit bullshitting on the side ;)
<orivej> gchristensen: it seems that darwin evaluations of nixpkgs:trunk are not going to ever get built. for example, https://hydra.nixos.org/eval/1467339 had 4285 queued jobs yesterday too
<LnL> orivej: only 2 of the macs are building currently so the darwin throughput is very low at the moment
<orivej> LnL: https://hydra.nixos.org/machines lists 4 active macs (all doing nixpkgs:staging), and it seems that low throughput does not explain no builds in 20+ hours
<LnL> no builds in 20+ hours?
<LnL> last darwin build finished 50s ago
<orivej> LnL: no nixpkgs:trunk darwin builds. I'm not saying that mac builders are doing nothing, only that nixpkgs:trunk darwin builds are not scheduled for a suspiciously long time
<LnL> ah
<LnL> that has to do with scheduling shares, or maybe somebody bumped the priority of staging because there's something important in there that should be merged
<orivej> it is possible that all is going as intended (although not due to higher shares), but it seems very likely that it is not
capisce has quit [Read error: Connection reset by peer]
capisce has joined #nixos-dev
terrorjack has quit []
terrorjack has joined #nixos-dev
orivej has quit [Ping timeout: 264 seconds]
<infinisil> clever: andi-: https://github.com/awakesecurity/ninja2nix was an experiment by taktoa[c]
woffs has quit [Quit: Gateway shutdown]
woffs has joined #nixos-dev
__Sander__ has joined #nixos-dev
ckauhaus has joined #nixos-dev
<clever> infinisil: yep, but he ran into issues with ninja generating the dep list at the same time as building
<clever> infinisil: so you cant get the deps until after you have built a thing
pie__ has quit [Ping timeout: 268 seconds]
<gchristensen> I'll spend some time this morning working on the macs
cocreature has joined #nixos-dev
<LnL> gchristensen: <3
orivej has joined #nixos-dev
<gchristensen> niksnut: I emailed everyone who hasn't ticked their name on https://github.com/NixOS/nixpkgs/issues/42761 saying it would be done at 1800 UTC today, this got a few more people to turn it on and one person replied saying they're okay with being removed
<{^_^}> #42761 (by grahamc, open): Require 2FA for all committers
<gchristensen> so I guess we're good to go any time after 1800 UTC
__Sander__ has quit [Quit: Konversation terminated!]
<niksnut> gchristensen: cool, thanks!
<gchristensen> delighted to have such good % so quickly add 2fa
<Profpatsch> gchristensen: You can find the mail of c0bw3b in the maintainers list.
<gchristensen> ohh
<Profpatsch> Not sure about mornfall, maybe they didn’t add their github handle?
<gchristensen> I think they contributed _ages_ ago
<Profpatsch> Ah, so they probably removed themselves from the maintainers list already.
<gchristensen> I mean
<Profpatsch> Also not active on Github at all it seems
<orivej> clever: actually my main use for fptrace is converting builds from their original build systems (make, ninja, whatever); this nicely avoids the need to parse makefiles and includes or the issue with ninja depfiles. (I have not tried converting them to Nix as ninja2nix does; that sounds funny to try.)
<clever> orivej: that sounds like it would need to do a non-incremental build once, to learn the deptree
<clever> and repeat that any time it changes
<orivej> clever: it needs one clean build, but why does it need to repeat it? Nix build will always be technically clean
<clever> orivej: if i change the source and add new deps, your old plan wont build them
<clever> orivej: what i was thinking of, is a tool that can parse the makefiles generated by configure, IFD them, then build everything in nix
<orivej> hm, ninja2nix seems good enough for this
<clever> ninja has a fatal feature that works similar to your plan
<clever> a step in the ninja file can generate its dep-list when building the file
<clever> so you wont know the deps until after it has built once
pie__ has joined #nixos-dev
<orivej> i see, neither approach will help with local development with a hypothetical nix-IFD-based build system
<gchristensen> or recursive nix
<clever> yeah, recursive nix could do it by just making a wrapper around gcc, and re-implementing ccache via nix
<clever> run the pre-processor, then just build a simple derivation that compiles it, and only depends on the hash of the preproccessed source
pie__ has quit [Ping timeout: 240 seconds]
<gchristensen> yay for checksums: sh: SHA-256 hash mismatch in 'https://nixos.org/releases/nix/nix-2.0.4/nix-2.0.4-x86_64-darwin.tar.bz2'; expected ec6279bb6d628867d82a6e751dac2bcb64ccea3194d753756a309f75fd704d4c, got ce2508efb8047ef9025b2c3fc33f6cd41ddf59b452a32a1eaf9e8d6676d46ec2
<gchristensen> FYI: I've erased and reinstalled Nix on mac2 and mac3 after their DBs were corrupted early this year, I think they haven't been working since. mac2 is back building and mac3 is almost ready to build. then, I'll be looking at mac6, mac7, and mac9 (ping LnL, orivej)
<shlevy> Wishlist: repl.nix that defines an attribute set, attributes are automatically put into scope when entering nix repl
orivej has quit [Ping timeout: 268 seconds]
<gchristensen> ohh ikwildrpepper / niksnut either of you have access to the macs any time soon? can't SSH to mac7 due to "No route to host"
<gchristensen> oops, mac6 not mac7 -- mac7 semes ok
Sonarpulse has joined #nixos-dev
<gchristensen> niksnut: min-free / max-free now enabled on the macs, so we'll get some good real-life testing on it
<infinisil> shlevy: So `nix repl repl.nix`?
<shlevy> oh does that already work? :D
<shlevy> amazing
<infinisil> Yup :)
<shlevy> Easiest feature request ever
jtojnar has quit [Remote host closed the connection]
<LnL> I have an alias for nix repl '<nixpkgs>'
jtojnar has joined #nixos-dev
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-dev
pie__ has joined #nixos-dev
Lisanna has quit [Ping timeout: 255 seconds]
orivej has joined #nixos-dev
Lisanna has joined #nixos-dev
<orivej> gchristensen: thank you, but this did not help unstuck nixpkgs:trunk (I expected that), there are as many queued jobs as before
<gchristensen> well that is a scheduling problem which should solve itself
<orivej> this looks more like a deadlock than like unfair scheduling, but ok
<LnL> scheduling with the shares can behave a bit unexpected with a large queue, it's not like trunk builds always win
<gchristensen> oh I see
bgamari has quit [Ping timeout: 240 seconds]
lopsided98 has quit [Ping timeout: 260 seconds]
bgamari has joined #nixos-dev
{^_^} has quit [Remote host closed the connection]
<gchristensen> ok yay, mac{1,2,3,4,5,7,8,9} are building again
bgamari has quit [Ping timeout: 256 seconds]
bgamari has joined #nixos-dev
<gchristensen> orivej: oh I see what you're looking at now, that is fishy
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
lopsided98 has joined #nixos-dev
bgamari has quit [Quit: ZNC 1.6.6 - http://znc.in]
bgamari has joined #nixos-dev
<gchristensen> orivej: nixpkgs:trunk jobs have fallen by about 100 in the last few minutes, so I think it is not deadlocked
<gchristensen> but perrhaps severely underprovisioned for a while.
<orivej> that sounds promising
<gchristensen> it'd be nice to be export these metrics in to prometheus
<gchristensen> I have two bug reports for Hydra
<gchristensen> but probably shoudln't make them public since they cause it to take a lot of resources
bgamari has quit [Quit: ZNC 1.6.6 - http://znc.in]
bgamari has joined #nixos-dev
orivej has quit [Ping timeout: 264 seconds]
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-dev
Sonarpulse has quit [Ping timeout: 255 seconds]
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
<gchristensen> niksnut: we're good for enabling 2FA whenever you're ready :)
<ikwildrpepper> yay
<niksnut> go for it!
<gchristensen> I can't do it, since I'm not an Owner :)
<niksnut> ok, will do once I'm finished installing my new laptop
<clever> niksnut: can you see anything obvious i'm doing wrong with this patch?, it causes nix to segfault: https://gist.github.com/cleverca22/d1fa877486493a7e3c40e72f859a8e0c
<gchristensen> ok, great!
<niksnut> gchristensen: I get "12 members will be removed"
<gchristensen> yeah
<gchristensen> I emailed them all and still
<LnL> most of the people that are left I've only seen active sporadically
<niksnut> okay, done
<gchristensen> yay!
<ikwildrpepper> so, now time to sign all commits :p
<ikwildrpepper> +require
* ikwildrpepper goes back into the troll-cave
<gchristensen> hehe
<gchristensen> ok niksnut you can use 147.75.96.102 for the bastion for me
<gchristensen> Packet.net lets you announce routes via BGP to "attach" elastic IPs :D
<clever> gchristensen: sounds like packet.net's peers arent doing proper filtering! :O
<gchristensen> they are
<clever> are you attaching an aws eip?
<gchristensen> its a carefully filtered and implemented feature
<gchristensen> no
<clever> ah
<dtz> so my hydra machine 's disk failed, and while I'm dealing with that (yay NixOS, but still) anyone know a quick way to "build like hydra" from the command-line? As in, I have a bunch of "release.nix"-style things that I'd like to manually build like Hydra would, and normal nix-build invocation isn't hydra-recursive (and needs to be told to "keep-going" :))?
<dtz> I seem to recall there was a way to do so but can't find it now :).
<gchristensen> niksnut: can you re-add cstrahan?
<LnL> dtz: recurseIntoAttrs or catching evaluation errors?
<dtz> I guess recurseIntoAttrs, although I suppose yes catching eval errors might be neded/useful too I forgot about that O:)
* dtz offers a sacrifice for "nix hydra-build" :P
<LnL> yeah, moving the hydra evaluator to nix would be nice
<LnL> so for errors it's not really possible AFAIK, builtins.tryEval catches less than hydra does
<LnL> but for the recursion you can write an expression :)
<dtz> okay, ty
orivej has joined #nixos-dev
<LnL> wait... I think nix-build already looks at recurseIntoAttrs
ckauhaus has quit [Quit: Leaving.]
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
phreedom has quit [Ping timeout: 250 seconds]
phreedom has joined #nixos-dev
contrapumpkin has quit [Remote host closed the connection]
genesis has quit [Read error: Connection reset by peer]
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
jtojnar has quit [Quit: jtojnar]
{^_^} has quit [Remote host closed the connection]
<rycee> Speaking of recurseIntoAttrs. Would somebody who knows about pkgs.newScope and Hydra mind having a look at https://github.com/NixOS/nixpkgs/pull/42659 ? It's about setting up perlPackages for different versions of Perl. It looks OK to me but I'm not familiar with how to best use pkgs.newScope and am a bit worried about the potential number of new packages introduced.
<gchristensen> why have so many versions of perl?
<rycee> I'm not sure. I'm not in a particular need of these, just having a look at the PR.
<LnL> we shouldn't recurse all of those
<rycee> Yeah, that was my feeling as well.
<LnL> and I'm not sure what size that is but for most other sets we only recurse on hydra, not with eg. nix-env
<rycee> My thought would be to just recurse into the primary one.
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
<gchristensen> why have so many versions of perl?
<gchristensen> oops
<gchristensen> why have so many versions of perl?
<gchristensen> .......
<gchristensen> why have so many versions of perl?
<gchristensen> OMG my client is freaking out, back later.
<LnL> you dislike perl _that_ much? ;p
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-dev
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-dev
{^_^} has quit [Changing host]
{^_^} has joined #nixos-dev
SuperKazuya has joined #nixos-dev