sphalerite changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | NixOS 19.03 released! https://discourse.nixos.org/t/nixos-19-03-release/2652 | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html https://r13y.com | 19.03 RMs: samueldr,sphalerite | https://logs.nix.samueldr.com/nixos-dev
niksnut has quit [Ping timeout: 272 seconds]
<samueldr> (thrice requeued)
niksnut has joined #nixos-dev
<gchristensen> :(
orivej has quit [Ping timeout: 252 seconds]
Synthetica has quit [Quit: Connection closed for inactivity]
alp has quit [Ping timeout: 258 seconds]
niksnut has quit [Ping timeout: 252 seconds]
niksnut has joined #nixos-dev
Drakonis has joined #nixos-dev
Drakonis has quit [Ping timeout: 258 seconds]
Guanin has quit [Remote host closed the connection]
<samueldr> ended up evaluating after another error evaluating
Drakonis has joined #nixos-dev
Drakonis has quit [Read error: Connection reset by peer]
noonien has quit [Read error: Connection reset by peer]
noonien has joined #nixos-dev
Synthetica has joined #nixos-dev
makefu has quit [Quit: WeeChat 2.4]
makefu has joined #nixos-dev
johanot has joined #nixos-dev
cjpbirkbeck has quit [Quit: Quitting now.]
alp has joined #nixos-dev
johanot has quit [Ping timeout: 272 seconds]
johanot has joined #nixos-dev
phreedom_ has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
Guanin has joined #nixos-dev
<teto> when writing tests, what's the diff between mustSucceed / succeed ?
<domenkozar[m]> any Nix news in last weeks? https://github.com/NixOS/nixos-weekly/pull/94
<{^_^}> nixos-weekly#94 (by domenkozar, 5 weeks ago, open): Call for Content: 2019/10
<domenkozar[m]> 5 :O
<niksnut> teto: they're the same
<niksnut> succeed is preferred
<teto> ok thanks
orivej has joined #nixos-dev
johanot has quit [Ping timeout: 248 seconds]
johanot has joined #nixos-dev
<{^_^}> #63766 (by domenkozar, 28 seconds ago, open): Add configurationLimit to systemd-boot to prevent running out of disk…
<domenkozar[m]> I'd really like to backport this
<domenkozar[m]> (can't rebuild my system without it :/ )
<Taneb> It seems reasonable
<domenkozar[m]> :grin:
_rvl has quit [Quit: ZNC 1.7.2 - https://znc.in]
_rvl has joined #nixos-dev
<niksnut> well, messing with the bootloader is *especially* something that shouldn't be done on the stable branch :p
<niksnut> because it's the one thing we can't recover from
<niksnut> if it goes wrong
<ekleog> OTOH looks like it's not changing anything for people without the option set
<domenkozar[m]> the only part that changes the bootloader runtime is `[-configurationLimit:]` which I've shown in the PR that if you substitute with 0, it's a noop
<domenkozar[m]> can also be made a bit more explicit
<domenkozar[m]> I'd prefer not to maintain a fork until september but I'm open to alternatives
init_6 has joined #nixos-dev
Jackneill has quit [Ping timeout: 245 seconds]
Jackneill has joined #nixos-dev
johanot has quit [Quit: WeeChat 2.4]
Jackneill has quit [Ping timeout: 258 seconds]
Jackneill has joined #nixos-dev
<manveru> domenkozar[m]: i'd love that for my rpi too :)
<gchristensen> I think it is a good idea too, we'll definitely need to get samueldr and sphalerite to look at it and weigh in w.r.t. backports, and maybe give it a few weeks on unstable before backporting
<niksnut> yeah a bit of soak time on master is a good idea
<domenkozar[m]> sgtm
<domenkozar[m]> thanks all :)
<domenkozar[m]> any thing that's missing for weekly tomorrow? https://github.com/NixOS/nixos-weekly/pull/94
<{^_^}> nixos-weekly#94 (by domenkozar, 5 weeks ago, open): Call for Content: 2019/10
<gchristensen> hmm probably would be good to publish about Snaps, but maybe only after snapcraft fixes some bugs in their system
orivej has quit [Ping timeout: 245 seconds]
<samueldr> manveru: if your rpi is using standard ext4, you can move the boot files to the main partition
<samueldr> (it is the new default in nixos-unstable since yesterday)
<samueldr> (well no, since the eval that started yesterday, not yet in unstable I think)
<manveru> heh
<manveru> well, i'm not gonna mess with it too much, since it sits in japan and i'd just like to keep it updated until i get back there
<samueldr> right, that would stink if you broke something along the way :/
<manveru> but yeah, should've just skipped making a boot partition, i was using nixos-generator for this
<samueldr> wait...
<samueldr> raspberry pi? systemd-boot?
<samueldr> oh, I thought you wanted *that* PR for the raspi, and not the same concept
<manveru> yeah
<samueldr> (though yest untested, now it should be possible to use systemd-boot with a small bit of setup)
<manveru> i'm using the `boot.loader.generic-extlinux-compatible`
init_6 has quit []
srhb has quit [Quit: ZNC 1.7.3 - https://znc.in]
srhb has joined #nixos-dev
<timokau[m]> gchristensen: Publish what about snaps?
<gchristensen> timokau[m]: my tooling to create snapcraft snaps
<timokau[m]> gchristensen: So not install them, but use `nix-build` to produce them? For any existing nix package?
<gchristensen> right
<gchristensen> in a few lines of code and a couple minutes of time. including firefox, which left the snapcraft team speechless
<timokau[m]> gchristensen: Sounds amazing, good work :)
<gchristensen> thanks :) PR is open, just waiting on upstream to fix a couple validation bugs
<domenkozar[m]> you can't just walk to a conference and solve their packaging problem
<domenkozar[m]> oh wait you can!
<gchristensen> :P
<andi-> I wonder why the docker people didn't receive it like that yet.. Maybe because snap is more suitable for Nix? I have no idea.
<ekleog> gchristensen: does that mean snap will start advertizing nix as a way to build snap packages? :D
<gchristensen> dunno :)
<gchristensen> let's get 1 or 2 useful snaps published before we start thinking that
<worldofpeace> jtojnar: looking at https://hydra.nixos.org/eval/1527051?compare=1525113#tabs-now-fail it looks fine
<worldofpeace> guess there was less than we anticipated.
<jtojnar> worldofpeace: I still see some weirdnesses
<jtojnar> https://hydra.nixos.org/build/95223261/nixlog/1 this is unrelated, we should probably drop it as valum is no longer developed AFAIR
<jtojnar> https://hydra.nixos.org/build/95200795/nixlog/1 also meson failure that looks unrelated
<jtojnar> zathura was expected as it fails the same way as girara did: https://hydra.nixos.org/build/95391566/nixlog/1
<worldofpeace> Yeah I've seen those except the one for gst-rtsp-server
<jtojnar> worldofpeace: the rtsp-server one was probably fixed by https://github.com/NixOS/nixpkgs/pull/54398/files#diff-52820f49214c07bb21ba3a85c92175d4
<worldofpeace> yeah it was jtojnar
<jtojnar> so rebase and merge?
<worldofpeace> I think so, not sure what else to check
pie_ has quit [Ping timeout: 268 seconds]
<worldofpeace> Wish I could expect all things to appear that simple 🙏
<jtojnar> worldofpeace: pushed (including the removal of auto-features from expressions)
<worldofpeace> jtojnar: looks good, cosmetically I think https://github.com/NixOS/nixpkgs/pull/63493/commits/58210dbb85ac4744131f6e2224996f28da382cd8 can be squashed into the prior
<jtojnar> worldofpeace: that prior one is cherry-picked from master
<jtojnar> perhaps master was not merged into staging yet
<jtojnar> will do that and then rebase again
<worldofpeace> yeah that makes sense
<jtojnar> ack conflicts
<jtojnar> I will just revert the latter, since the former is superior
<worldofpeace> that's pretty common with python packages, it moves so quick in nixpkgs
<ekleog> do we really want to have files specialized for python2 support?
<ekleog> like, python2 countdown is going to be over in a few months, not supporting peripheral packages like scikit over there would make sense to me (now, the work has already been done, so I'm not debating that, just wondering whether I should care about python2)
<gchristensen> we won't be able to just drop py2 at the EoL unfortunately
<gchristensen> so I think it is premature to start being too aggressive about cutting it away
<ekleog> definitely not, but we might be able to start phasing it out :)
<ekleog> like, not trying to keep support for packages whose latest version doesn't support python2 and which aren't needed for our own code to work
<ekleog> that said I guess timokau[m] will have a different view on the matter?
<ekleog> (my own view being, for the time being, “well… to drop py2 we need to start somewhere”)
cjpbirkbeck has joined #nixos-dev
Guanin has left #nixos-dev ["Leaving"]
Guanin has joined #nixos-dev
Guanin is now known as Amarandus
<jtojnar> worldofpeace: what do you think about the discource announcement?
<jtojnar> will post it here in order not to spam in the issue https://paste.gnome.org/ptdzk1t6n
<jtojnar> s/issue/PR/
<worldofpeace> jtojnar: replied on github
<jtojnar> worldofpeace: ok, will post it
<jtojnar> agree with you 100%
<worldofpeace> Maybe there should be a development announcements section?
<jtojnar> worldofpeace: yeah, just wanted to suggest that
<jtojnar> zimbatm^
orivej has joined #nixos-dev
<worldofpeace> Or maybe there should be development highlights each month, with important notices to maintainers too.
<jtojnar> that feels like too much work
<jtojnar> important changes announce list where people can just post something like a PR link with short description is lower effort
<worldofpeace> that's pretty much what I meant
<jtojnar> highlights mean that some curator produces them (like nixos-weekly), I would prefer something more organic where anyone can post and things are immediately published
<jtojnar> s/mean/mean to me/
<samueldr> yeah, a discourse category "just so you know I did ______"
pie_ has joined #nixos-dev
<samueldr> that also may help RMs figuring out changes for the next release's release notes
<worldofpeace> yeah and discourse already would work perfect for that
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
<worldofpeace> samueldr: getting release notes together for a nixos release sounds like a lot of work
<samueldr> it's not as bad as it looks, it's mostly a long winded effort to comb through commits/PRs rather than a complicated task requiring thinking
<samueldr> and with git, checking the history of only one of the base folder at a time makes it easier, check lib for changes, check pkgs for notable changes, check nixos for notable changes
<samueldr> with --merges-only and --no-merges (iirc) it also makes it easier to understand
<worldofpeace> I thought thinking was easy for humans and sorting was for machines :P
<worldofpeace> but I guess we adapt
<samueldr> just enough thinking so it's harder for machines :), though I think I also grepped away version bumps
<worldofpeace> is the `merges-only` part documented?
<samueldr> in our docs, no, I think there's no tips for figuring out the release notes
<samueldr> it's `--merges` and `--no-merges` for git log, tig also respects those parameters
<worldofpeace> just asking because there might be a potential future release manager mentally acclimatizing 😸
* gchristensen perks up
<worldofpeace> but they're pretty unsure how they'll handle it
<jtojnar> worldofpeace: I just filed it under Uncategorized for now https://discourse.nixos.org/t/psa-meson-now-building-all-auto-detected-features/3271
<worldofpeace> jtojnar: nice.
<worldofpeace> my meson itch is now pulling me torwards #56199
<{^_^}> https://github.com/NixOS/nixpkgs/pull/56199 (by eadwu, 17 weeks ago, open): mesa: build with meson
<timokau[m]> ekleog: I'll do a little happy dance as soon as python2 dies
<timokau[m]> I don't have a very strong opinion on weather or not we want to keep py2 support for scikitlearn. Just by default I try not to unnecessarily break things for people
<ekleog> 'k )
<ekleog> :)*
<timokau[m]> gchristensen: That question may be naive, but why won't we be able to drop support at EOL?
<gchristensen> because it is impossible, far too many very important things depend on py2
<gchristensen> effectively no (widely used) distro is able to do this
<timokau[m]> Hm, that sounds like we'll never get rid of python2 :/
<timokau[m]> Do you have examples for Very Important Things that still only support py2?
<gchristensen> I feel quite sick right now, so I'm not going to go looking too hard
<samueldr> could python2 be isolated, so that it doesn't hold back things incompatible with it in the infra, while allowing to use it as a dependency for those important bits in the time?
<timokau[m]> Doesn't matter if you don't know anything f the top of your head, just idle curiosity :)
<gchristensen> samueldr: that might be
<samueldr> some kind of pkgs.__warrantyVoidIfUsed.python2 :)
<niksnut> meta.broken?
<samueldr> wouldn't that make transitive users broken too?
<niksnut> yes
alp has quit [Ping timeout: 258 seconds]
alp has joined #nixos-dev
Drakonis has joined #nixos-dev
pie_ has quit [Remote host closed the connection]
cjpbirkbeck has quit [Quit: Quitting now.]
pie_ has joined #nixos-dev
Drakonis has quit [Read error: Connection reset by peer]
pie_ has quit [Ping timeout: 268 seconds]
pie_ has joined #nixos-dev
matthewbauer has joined #nixos-dev
<matthewbauer> niksnut: Would it be possible to integrate `nix-bundle` functionality into Nix? Right now it still uses the old-style Nix commands, but I wonder if it deserves its own commands
<matthewbauer> niksnut: I am not sure if it makes sense, nix-bundle is still enough of a hack to make it maybe not worth supporting 100%. However, guix has its pack command now: https://www.gnu.org/software/guix/manual/en/html_node/Invoking-guix-pack.html which does a similar thing
<matthewbauer> This was brought up in https://github.com/matthewbauer/nix-bundle/issues/47
<{^_^}> matthewbauer/nix-bundle#47 (by haslersn, 1 day ago, open): Make CLI similar to nix build
<niksnut> matthewbauer: I think that would be pretty cool
<tilpner> niksnut: What about other subcommands, like generators for expression (node, cabal, cargo -> nix)? guix integrated those too
<niksnut> could be, but they'd have to be written in C++ (or maybe rust in the future)
<niksnut> what dependencies would be needed for nix bundle?
<qyliss> I wouldn't like that I think. The best way to generate those expressions changes very regularly, and experimentation and there not being one "blessed" way is pretty important.
<niksnut> yeah
<ekleog> OTOH, having one “blessed” way is a good way to say “this is supported” -- personally, with the plethora of python to nix things, I'm completely lost
<ekleog> The “blessed” way can still evolve, it's not set in stone
<ekleog> so long as the overall API is fixed, that is
<qyliss> turnaround time is way bigger with stuff in Nix core, and it's hard enough to experiment with new stuff as it is.
<qyliss> As a Bundix maintainer, it would be a lot more difficult for me to iterate on Ruby packaging if I had to go through Nix
<ekleog> You wouldn't have to go through nix
<ekleog> you could maintain bundix, and there to be a nix way that'd be updated once in a while to the latest bundix
<ekleog> now, forcing things to be written in C++/rust is unfortunate, OTOH, and would likely kill such attempts anyway
<samueldr> nixos:trunk-combined, Too many heap sections: Increase MAXHINCR or MAX_HEAP_SECTS ~ 14:27 UTC
averell has quit [Excess Flood]
averell has joined #nixos-dev
das_j has quit [Quit: "Bye!";]
pie_ has quit [Ping timeout: 244 seconds]
das_j has joined #nixos-dev
tilpner has quit [Quit: WeeChat 2.4]
<gchristensen> I propose we give flokli merge rights
<gchristensen> oh he already has it (thought he didn't)
<andi-> hu? I think that has been the case since february
<andi-> or longer
<samueldr> I guess no one objects then :)
<gchristensen> :P
<flokli> :-D
<flokli> thanks for the reassurement, I guess ;-)
pie_ has joined #nixos-dev
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
alp has quit [Ping timeout: 259 seconds]
Amarandus has quit [Remote host closed the connection]
Synthetica has quit [Quit: Connection closed for inactivity]
cjpbirkbeck has joined #nixos-dev