zimbatm changed the topic of #nixcon to: NixCon is live! https://www.youtube.com/c/NixCon/ - 25-27 October 2018 - London - https://nixos2018.org , organisation is on https://trello.com/nixcon
Taneb has quit [Quit: I seem to have stopped.]
Taneb has joined #nixcon
lassulus_ has joined #nixcon
lassulus has quit [Ping timeout: 246 seconds]
lassulus_ is now known as lassulus
jasongrossman has quit [Ping timeout: 240 seconds]
jasongrossman has joined #nixcon
worldofpeace has quit [Ping timeout: 250 seconds]
worldofpeace has joined #nixcon
worldofpeace has quit [Ping timeout: 264 seconds]
ryantm_ has joined #nixcon
ivan_ has joined #nixcon
arianvp_ has joined #nixcon
arianvp has quit [*.net *.split]
domenkozar has quit [*.net *.split]
ivan has quit [*.net *.split]
domenkozar has joined #nixcon
FRidh has joined #nixcon
FRidh has quit [Client Quit]
<andi-> Another beautiful Nixcon day \o/
clever_android has quit [Read error: Connection reset by peer]
clever_android has joined #nixcon
TanebPhone has joined #nixcon
<TanebPhone> I've just realised I left my badge at home, can I get in without it?
<srhb> TanebPhone: Probably needs a zimbatm ping :)
<TanebPhone> zimbatm: ^^
<layus[m]> You will certainly have too look at the conference through the Windows now ;-)
<TanebPhone> :C
<zimbatm> TanebPhone: just come, we'll hand you a blank badge
<TanebPhone> Thanks! I'm sorry for the inconvenience
TanebPhone has quit [Quit: AndroIRC - Android IRC Client ( http://www.androirc.com )]
<zimbatm> no problem
<zimbatm> adisbladis: goibhniu[m] : flokli : are you guys hungover? :D
clever_android has quit [Ping timeout: 276 seconds]
<andi-> zimbatm: almost there ;)
clever_android has joined #nixcon
clever_android has quit [Ping timeout: 246 seconds]
clever_android has joined #nixcon
clever_android has quit [Ping timeout: 272 seconds]
clever_android has joined #nixcon
<adisbladis[m]1> I did not sleep well last night.. Almost at the venue :)
goibhniu has joined #nixcon
<clever> goibhniu: and your live!
zarel has joined #nixcon
<goibhniu> \ó/
<shlevy> For some reason that guy looks balder than the normal \o/
<gchristensen> omg that is amazing
kriztw has joined #nixcon
<clever> gchristensen: yeah, i can see how this would make nixops a lot more extensible, with customizable backends
<srhb> goibhniu: Is it just me or is audio and video out of sync?
<srhb> I don't think this was the case yesterday.
<srhb> (No biggie)
<goibhniu> oh! thanks! ... can someone confirm?
<clever> its on mute on my end, since i'm in the room
<srhb> I think video is like 1 sec behind audio. No problem, just a bit weird.
<goibhniu> ah, that's quite bad
<kriztw> less than a second, but I noticed it too
<clever> ive noticed that the screen capture colors are a bit off, the whites arent white
<srhb> goibhniu: Hours of resyncing awaits... :-)
* srhb has been there with unmetronomed systems D:
<clever> srhb: one of those clappers from movies might help? lol
<andi-> It sounds like a real nice approach.. Should give it a spin soon™
<srhb> clever: I prefer having a digital time source, but yes. xD
<clever> srhb: linus tech tips has something that inserts a digital timecode into the audio stream, but i'm not sure how exactly it works
<goibhniu> I'll try restarting all the things at the break ... the settings haven't changed since yesterday though ... hrm
<nbp> adisbladis: Question for the audience: In the room, how many persons are interested in having modules which can only change a limited set of option?
<srhb> goibhniu: Ok, I'll report back after the break :)
<gchristensen> nbp: :hand:
<goibhniu> yeah, that's weird about the colors too ... yesterday morning they seemed really quite bad, and got better during the day
<nbp> my point is that I can make something quickly/ugly which can answer this question.
<adisbladis> nbp: Huh, how can the room answer? ;)
<srhb> adisbladis: count the hands :-P
<nbp> adisbladis: raising hands, and rough estimate.
<nbp> Hum … I guess one can use submodules as a way to make a public interface.
<nbp> ^ without changing the current module system.
<nbp> ^ which also solve the namespace issue.
<infinisil> nbp: not entirely, you can't use imports with submodules
<nbp> infinisil: You can.
<infinisil> Yet at least, I've been trying to get this to work
<nbp> infinisil: if not this should be easy to fix.
<infinisil> You can but there are some problems with it
<srhb> Today in Nixcon: Speakers with varyingly named Richard.
<infinisil> Hehe
ivan_ is now known as ivan
<goibhniu> srhb: still out of sync, right?
<srhb> goibhniu: Yes
<goibhniu> cheers!
xeji has joined #nixcon
johanot has joined #nixcon
* samueldr loves 01-jan-1970 directory listings :)
<arianvp_> morning
arianvp_ is now known as arianvp
<arianvp> catching up at 2x speed
<srhb> arianvp: See you in about an hour then xD
<arianvp> in theory with all the breaks included this should work
<arianvp> :')
xeji has quit [Ping timeout: 246 seconds]
zarel has quit [Ping timeout: 264 seconds]
jasongro` has joined #nixcon
zarel has joined #nixcon
jasongrossman has quit [Ping timeout: 252 seconds]
<goibhniu> ... and we're back!
<goibhniu> lemme know if the sync is any better
<sphalerite> \ö/
<srhb> goibhniu: It's better. Not as good as yesterday. :)
<srhb> (Sound is still slightly ahead)
<goibhniu> hrm ... thanks for checking!
<goibhniu> the recordings are also out of sync
<srhb> Yup.
<srhb> I'm surprised that it was this good yday if you're not using any sort of sync source. :)
<srhb> that good*
xeji has joined #nixcon
<srhb> (It's really not that bad. :-))
<srhb> I'm just happy I get to see the con :D
<goibhniu> \o/
<goibhniu> excellent mic technique!
<srhb> Very! :D
<srhb> There should be a prize
<arianvp> what a pro
<arianvp> both mic-wise and mips-wise
<srhb> *crisp, full voice* is this volume level ok for everyone
<srhb> troll. :PP
<clever> somebody from #nixos had trouble running multi-user nix in docker yesterday, due to nix-daemon not persisting as root
<yorick> last time I tried to install nix I found out you can't install multi-user nix if you're root
<gchristensen> yeah... that is silly
<infinisil> #43795
<infinisil> Ohh bot not here
zarel has quit [Remote host closed the connection]
johanot has quit [Ping timeout: 245 seconds]
xeji has quit [Ping timeout: 240 seconds]
aanderse-work has joined #nixcon
aanderse-work has quit [Quit: Leaving.]
johanot has joined #nixcon
zarel has joined #nixcon
zarel has quit [Ping timeout: 264 seconds]
zarel has joined #nixcon
xeji has joined #nixcon
zarel has quit [Ping timeout: 272 seconds]
pie__ has joined #nixcon
xeji has quit [Quit: WeeChat 2.2]
jasongro` has quit [Quit: ERC (IRC client for Emacs 26.1)]
drakonis has joined #nixcon
jasongrossman has joined #nixcon
xeji has joined #nixcon
<disasm> fpletz: are you manually writing those networkd files (since you said some features aren't supported in nix) or do you have nix scripting that's not upstream that generates it?
<fpletz> disasm: not upstream unfortunately… the PR will be ready soon :)
<samueldr> fpletz: remember to do the work weeeeell before february for a 19.03 release!
<fpletz> samueldr: I'll try! :)
<clever> fpletz: the routing table stuff sounds interesting for solving some issues with toxvpn
xeji has quit [Ping timeout: 264 seconds]
<clever> fpletz: basically i have no way of knowing what IP's toxvpn contacts at runtime, so i need to give that process full internet access
<clever> fpletz: and that makes it difficult to default all traffic into the vpn
<samueldr> and fpletz, as far as I'm concerned, seems pretty cromulent to change as a default
<clever> fpletz: being able to have seperate routing tables for the vpn client, and the rest of the os, would make it trivial to vpn everything over toxvpn
<cransom> there is no shortage of silly tricks you can do in linux to maniuplate traffic, but vrfs are good.
TanebPhone has joined #nixcon
<clever> cransom: without vrfs, i would somehow need to transfer wlp3s0 to a network namespace where i run the vpn, and then setup a normal container bridge between the 2 namespaces, to expose the gateway IF
<clever> oh, that first part seems easy, `ip link set eth10 netns myns` i think
johanot has quit [Ping timeout: 246 seconds]
<cransom> you can also setns a process that's in a table that would have your default route
johanot has joined #nixcon
zarel has joined #nixcon
emily has quit [Ping timeout: 276 seconds]
TanebPhone has quit [Quit: AndroIRC - Android IRC Client ( http://www.androirc.com )]
xeji has joined #nixcon
zarel has quit [Ping timeout: 252 seconds]
drakonis has quit [Quit: WeeChat 2.2]
<ckauhaus> fpletz: I think the networking stuff is complex enough to warrant a hackday or two - perhaps in Munich?
<ckauhaus> we could try to get 3-4 people together and get stuff done
<ckauhaus> I'd be interested
xeji has quit [Ping timeout: 246 seconds]
xeji has joined #nixcon
<fpletz> clever: exactly :) it's also easier to set up than network namespaces (you don't have to use ip netns exec ip …)
<fpletz> also, if you bind sockets in services to the vrf interface, their traffic will be routed via the associated routing table
<clever> fpletz: i dont have full control over the socket creation in my vpn
<clever> fpletz: i'm reusing an IM library, toxcore, for the network layer, and its p2p
<fpletz> ckauhaus: yeah, let's do that! :) mic92 and andi- will probably also be interested
<arianvp> me too!
<arianvp> I'm in munich 15 november :P
<ckauhaus> November would be ok, or do it in January
<fpletz> as far as timing is conncerned, that doesn't sound unreasonable
<arianvp> I and (I think andi- too) are at Munihac around that time
<fpletz> oh, right, I'll also be there
<arianvp> oh cool
<fpletz> perfect timing
<fpletz> I'll organize someting
<arianvp> I arrive the thursday before
<arianvp> but arrive in the evening though.
<andi-> I arrive at around midnight that day :D
<andi-> so only Friday, Sat- & Sunday
<arianvp> perfect. so no sleep
<andi-> but those where actually already booked (at least day time)
<ckauhaus> well
<andi-> but we can do that during the night ;-)
<ckauhaus> does not necessarily make the code better
<andi-> or worse :P
<fpletz> we have enough flora power at mayflower :>
<fpletz> to keep you awake
<adisbladis> Networking AND flora... Appealing!
<ckauhaus> I'd prefer to do it when we have our heads clear
<fpletz> sure, just kidding :)
<ckauhaus> so better go for the Monday after?
<andi-> Then lets pick a weekend ? (Friday evening -> Sunday afternoon)
<ckauhaus> I could free up a day during the week, not sure if everyone is so flexible
<arianvp> I'm not flexible but that's ok. you can also do it without me, can brainstorm a bit in IRC
<andi-> My current job isn't flexible enough for that... They don't value nix (yet) ;-)
<arianvp> unemployed is underrated \o/
<ckauhaus> lol
<fpletz> I'll create a dudle with some dates and let's see what date fits best
<ckauhaus> yeah
<fpletz> but we can talk and hack a little around munihac anyway
<andi-> yep
<arianvp> is Sander in this chat? I had a question about his talk
<ckauhaus> perhaps set a specific day/time aside - I'm not a Haskeller so I would only come for networkd hacking not for the rest
<adisbladis> arianvp: Ask away
<andi-> arianvp: ping adisbladis with the question?
<adisbladis> I'll relay it to the microphone
* samueldr was ninja'd
<ckauhaus> "the Internet has a question" *g*
<arianvp> How are migrations handled?
<arianvp> for say a your mysql db
<gchristensen> oh boy
<arianvp> as in, schema migrations. not location migrations
<arianvp> About the key question: Vault is exactly a database for keys! perhaps we could manage it in dysnomia :)
<arianvp> Thanks for relaying the question! :D
<adisbladis> Np! :)
<pie__> fpletz, is it possible to use networkd for a simple configuration thats just waiting for a vpn device to come up and then assigning it an ip/route, with whatever is currently in nixpkgs?
worldofpeace has joined #nixcon
<gchristensen> you hear that? we have good documentation!
<pie__> gchristensen, ;P
xeji has quit [Ping timeout: 246 seconds]
<flokli> pie__: you can drop a .network file matching on a specific name, mac addr driver and more (or combinations of that) - networkd doesn't necessarily needs to have set up that device before
qyliss has quit [Quit: bye]
qyliss^work has quit [Quit: bye]
Shell has quit [Quit: ZNC 1.6.5 - http://znc.in]
xeji has joined #nixcon
qyliss^work has joined #nixcon
<nh2> fpletz: what's the plan for integrating the networkd switch with downstream tools? If you want to make it default for 19.03, how do we make sure that e.g. nixops is fully compatible with it, if anything needs to be changed? Or (assuming that there'll be an option to switch to the old scripts until they are removed) would we tell nixops users to use that switch until some later point?
qyliss has joined #nixcon
<ckauhaus> Honestly, I don't think it will already be default for 19.03
<ckauhaus> The goal should be to make it a viable option for 19.03 and perhaps switch the default on 19.09
<ckauhaus> that stuff runs too deep into the system
* cransom will flip it on as soon as he's in front of the router again. i'd rather not chop it's legs off and upset the natives.
xeji has quit [Ping timeout: 246 seconds]
<gchristensen> +1
infinisil has quit [Quit: Configuring ZNC, sorry for the joins/quits!]
<gchristensen> though I am very tempted to try it ~today on my Packet machines
<ckauhaus> heh
infinisil has joined #nixcon
infinisil has quit [Client Quit]
infinisil has joined #nixcon
xeji has joined #nixcon
* fpletz is feeling so bad he didn't finish his PR in time for nixcon… with that the experience will be much smoother
<fpletz> nh2: yeah, haven't haven't checked if upgrades will work smoothly. we'll sort all that out soon :)
<ckauhaus> fpletz: could you give me a pointer to your nginx reload PR?
<ckauhaus> thx
Shell has joined #nixcon
<disasm> FYI, I've had a lot of people ask me where my nix configs are/if my slides are available: https://github.com/disassembler/network https://util.samleathers.com/presentations/nix-at-home.html
qyliss^work has quit [Quit: bye]
qyliss has quit [Quit: bye]
Shell has quit [Quit: ZNC 1.7.1 - https://znc.in]
<arianvp> oh man this talk
<arianvp> i'm just nodding in agreement all the time
<arianvp> adisbladis: Q from the internet for later: Why do the lockfiles not contain the hash of the nix file that is being imported?
<nbp> If we go that way, the Nix performance issue people are reporting on NixOS is going to be much worse in a near future.
<arianvp> so this would be an 'evaluation-time' import then?
<adisbladis> arianvp: Ack :)
<nbp> arianvp: no, a memory issue, which unfortunately cannot be solved easily
<clever> shlevy: there is cleanSourceWith, that remembers the functions and original path, and will combine filters together
<arianvp> this seems very similar to Eelco's idea about modules as well right?
<ckauhaus> there's at least quite a bit of overlap
<nbp> I love documentation examples, which are always more documented than what developers do usually.
<nbp> ^ foo.enable.description = "Enable foo.";
<pie__> yeah but what are you studying secifically xP
<pie__> ^ wrong chan
<arianvp> I think this talk should be taking into account in the package module system RFC :)
<arianvp> awesome stuff
qyliss has joined #nixcon
qyliss^work has joined #nixcon
<nbp> arianvp: I think this talk make sense if we want to get rid of Nixpkgs as a central repository
<nbp> arianvp: and move Nix expressions to each package.
<nbp> arianvp: However, I fear this would not work well with security quality requirements.
<clever> nbp: any IFD at all in nixpkgs will destroy the performance of nix-env -i
qyliss has quit [Client Quit]
qyliss^work has quit [Client Quit]
<nbp> clever: IFD ?
<clever> nbp: import from derivation
<arianvp> monorepos do lend themselves well for security updates, indeed.
<arianvp> clever: I think the idea is to use builtins.fetchurl here
<clever> nbp: nix-env -i will scan the .name of every derivation
<arianvp> so at evaluation time
<nbp> clever: IFD is a gargantuan NO.
<adisbladis> arianvp: Having a "metarepo" could solve that problem even better
<clever> arianvp: even if your using fetchurl on a nix file, it will have to download 100's of nix files before nix-env -i hello can find it
<adisbladis> Ie not blocking a curl update because the kde desktop is broken
<arianvp> yes, but maybe some meta-package which is a list of package names paired with their fetchurl is needed then
<arianvp> a "package registry" file
<snajpa> yo guys, regarding tomorrow's hackaton, seeing how hard core guys are here... if you tell me what you want to hack on, I'll give you my ticket, if it's interesting
<arianvp> nix-env -i could consume that
<clever> arianvp: attrpaths are already that
<arianvp> yes exactly.
<clever> arianvp: nix-env -iA nixos.hello wont have the issue
<nbp> adisbladis: Q: How much time of any nix command depends on fetching require-s from the network?
<arianvp> isn't the idea to get rid of nix-env -i in nix 2.0 anyway? and do everything attr based?
Shell has joined #nixcon
qyliss has joined #nixcon
<arianvp> in nix-ng*
<clever> arianvp: probably, i would support that change
qyliss^work has joined #nixcon
<ckauhaus> nix 2.0 is always here and still supports -i ;-)
<nbp> adisbladis: reworded-Q: How frequently do we have to fetch from the network when using require.nix style?
<arianvp> I mean the new nix commands living under "nix *"
<clever> arianvp: another pain-point is that things like hydra would still lag a lot, it blocks when fetching a url, and cant eval other things during that time
<arianvp> there isn't an "env" equivalent yet
<nbp> adisbladis: (thanks, forget the reworded one then)
<snajpa> if you know anyone who'd use the hackaton ticket better than I will... pls let them know Im offering it :)
<arianvp> clever: This might mean CI should be distributed as well? (Not sure if this is a good answer :P )
<ckauhaus> aside: Anyone out for a pub tonight?
* infinisil lifts hand
<nbp> I think a part of this work would make sense if the fetching is not checked every time.
<nbp> But only done on an explicit fetch command.
<clever> nbp: if you supply a sha256 to the fetch, it can cache and never retry
<nbp> clever: no because of the network accesses.
<nbp> clever: Have you tried using the rust-overlay when you are on a crappy network connection?
<clever> nope
<nbp> clever: then don't, you just either wait to download something which never ends, or wait 5 minutes until you get the last one in cache.
xeji has quit [Ping timeout: 252 seconds]
xeji has joined #nixcon
<manveru> ckauhaus: me too
<ckauhaus> no idea where to go, though :-/
<ckauhaus> any local around?
<nbp> ckauhaus: ask zimbatm
<manveru> I think the once we went to the first day was nice
<manveru> But it's super cold :(
<ckauhaus> perhaps discuss the question f2f right now
johanot has quit [Quit: WeeChat 2.2]
xeji has quit [Ping timeout: 245 seconds]
goibhniu has quit [Ping timeout: 246 seconds]
<ckauhaus> some ppl will meet at 7pm in front of the Southwark tube stn and look out for a pub nearby
nikivi has quit [Quit: ZNC is awesome]
clever_android has quit [Ping timeout: 240 seconds]
clever_android has joined #nixcon
nikivi has joined #nixcon
moredhel has joined #nixcon
<flokli> ckauhaus: what about going to the same pub as yesterday? I liked some of the beers there, and the view to the river ;-)
clever_android has quit [Read error: Connection reset by peer]
clever_android has joined #nixcon
pie__ has quit [Ping timeout: 272 seconds]
pie_ has joined #nixcon
zarel has joined #nixcon
pie_ has quit [Remote host closed the connection]
zarel has quit [Quit: Leaving]
clever_android has quit [Ping timeout: 252 seconds]
<disasm> ckauhaus: where is everyone at? Still drinking somewhere? Love to join you guys!
<adisbladis[m]1> Disasm: ^
<adisbladis[m]1> We're about 10 nixers hanging out
<tazjin> about 10 more at brewdog shoreditch
clever_android has joined #nixcon
clever_android has quit [Ping timeout: 264 seconds]
clever_android has joined #nixcon
clever has quit [Ping timeout: 260 seconds]