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
orivej has quit [Ping timeout: 264 seconds]
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 256 seconds]
lassulus_ is now known as lassulus
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
orivej has joined #nixos-dev
vcunat has joined #nixos-dev
vcunat has quit [Client Quit]
pie_ has quit [Ping timeout: 248 seconds]
infinisi1 has joined #nixos-dev
infinisil has quit [Read error: Connection reset by peer]
FRidh has joined #nixos-dev
orivej has quit [Ping timeout: 260 seconds]
__Sander__ has joined #nixos-dev
sir_guy_carleton has quit [Quit: WeeChat 2.0]
goibhniu has joined #nixos-dev
ben has quit [Ping timeout: 255 seconds]
orivej has joined #nixos-dev
phreedom has quit [Ping timeout: 250 seconds]
phreedom has joined #nixos-dev
ben______ has joined #nixos-dev
orivej has quit [Ping timeout: 256 seconds]
the has joined #nixos-dev
the has joined #nixos-dev
the has quit [Changing host]
the has quit [Remote host closed the connection]
the has joined #nixos-dev
the has joined #nixos-dev
the has quit [Changing host]
__Sander__ has quit [Quit: Konversation terminated!]
<gchristensen> infinisi1: ping
<infinisi1> gchristensen: pong
<gchristensen> infinisi1: how do I tell {^_^} to ping me?
<gchristensen> oops
<gchristensen> infinisi1: how do I tell {^_^} to ping me when vcunat is around
<infinisi1> ,tell should work
<{^_^}> infinisi1: I'll pass that on to should
<infinisi1> ,tell
<{^_^}> infinisi1: Use `,tell john Remember to do the laundry` to send this to john next time he's talking in this channel
<infinisi1> And it mentions your nick in that message so you should get notified too
<gchristensen> ,tell vcunat please ping me (gchristensen) when you're around, w.r.t. your cache slowdown. Please run `curl -I https://cache.nixos.org` a few times, and send me the results (headers specifically)
<{^_^}> gchristensen: I'll pass that on to vcunat
should has joined #nixos-dev
<gchristensen> thanks, {^_^}
<should> oh right, it doesn't trigger on join
<{^_^}> should: 2 minutes, 27 seconds ago <infinisi1> work
<samueldr> (it was me)
<infinisi1> Yeah, i don't even have access to the joins from the frontend
should has quit [Ping timeout: 252 seconds]
Sonarpulse has joined #nixos-dev
<aminechikhaoui> gchristensen: btw I think the script already had some curl output with the cloudfront headers ?
<gchristensen> seems they wanted more traces?
<gchristensen> did I misunderstand?
<aminechikhaoui> yeah maybe they needed more, won't hurt anyway :)
<gchristensen> can update the cloudfront.sh to include it :)
<gchristensen> ,tell vcunat I updated https://gist.github.com/grahamc/df1bb806eb3552650d03eef7036a72ba if you want to just re-run that script or whatever :P
<{^_^}> gchristensen: I'll pass that on to vcunat
<gchristensen> thanks {^_^}
ben______ is now known as ben
ckauhaus has quit [Quit: WeeChat 2.0]
Sonarpulse has quit [Ping timeout: 260 seconds]
<Profpatsch> gchristensen: On freenode there’s also MemoServ
the has quit [Excess Flood]
orivej has joined #nixos-dev
the has joined #nixos-dev
the has joined #nixos-dev
the has quit [Changing host]
<gchristensen> one patch at a time, the becomes a better and better ban machine
pie_ has joined #nixos-dev
<andi-> does it learn patterns on it's own yet?
<gchristensen> no
<andi-> is there a reason it only does nickname bans vs hosts?
<gchristensen> I hadn't thought about it
FRidh has quit [Quit: Konversation terminated!]
<gchristensen> some day I'll probably update it to go in and clear the ban-lists for all our channels since botnet sufferers don't really need to be banned forever
peti has quit [Quit: vacation]
Sonarpulse has joined #nixos-dev
<Profpatsch> Unpopular opinion:
<Profpatsch> stdenv should fail if any files remain in . after the build.
<gchristensen> O.o
<simpson> Profpatsch: Ha, fun idea.
<Profpatsch> We need better linters, better introspection mechanisms.
<Profpatsch> Better builds of docs.
<Profpatsch> Each package in the core set must contain a manpage or some form of doc
<Profpatsch> The license files, changelogs and other files should be accessible in the build outputs.
<Profpatsch> And if possible machine-indexed.
<samueldr> Profpatsch++
<{^_^}> Profpatsch's karma got increased to 1
<samueldr> though, one big flaw I foresee: cargo culting rm -rf * into builds as last steps
<gchristensen> the rm -rf seems like a low-value thing
<Profpatsch> samueldr: At least you can see that something is wrong this way.
<gchristensen> the rest of it seems great
<gchristensen> I see nothing wrong with stuff in . indicating a bad thing
<gchristensen> I see nothing wrong with stuff in . *
<Profpatsch> Maybe saying “nothing left in .” is too indirect. The idea is: Nothing unspecified.
<Profpatsch> Specified also means: Throwing it away explicitely if you don’t want to keep it in the output.
<Profpatsch> e.g. by having a list of globs of what you don’t want to keep.
<gchristensen> this sounds like making work to add work. where do you see the value?
<Profpatsch> And checking that . is empty is a sanity check.
<Profpatsch> Right now a lot of pages are missing manpages, leaving folders with docs in . instead of copying them to man.
<Profpatsch> Or AUTHORS files, Changelogs …
<Profpatsch> Stuff that’s valuable to have in the output.
<Profpatsch> It’s similar in value to splitting outputs I’d say.
<Profpatsch> Or having checks that outputs don’t reference each other if possible.
<gchristensen> I disagree, these are tasks maintainers should be doing, and adding busywork of rm -rf ./* isn't really fixing it
<Profpatsch> Well, rm -rfing shouldn’t be able to go into the package set.
<gchristensen> yes: add the expectation (and verify) that docs get populated, yes: copy over changelogs and stuff, yes: automate parts that are really common in to stdenv, no: I think it is unreasonable to expect the build workspace to be in some very specific state post build
<Profpatsch> It also helps if new versions introduce new files or move them
pie_ has quit [Read error: Connection reset by peer]
pie__ has joined #nixos-dev
<samueldr> I understand the intention behind all that, but wow that's busywork :(
pie__ has quit [Remote host closed the connection]
<samueldr> unless there is tooling that undestands common things, or cleans up make clean make distclean, remove initial source files
<samueldr> (because AFAIK nothing enforces out of tree builds)
<samueldr> and then, even removing initial source files is prone to not copying authors, changelogs, etc.
<gchristensen> and make clean will delete the generated docs
<samueldr> oh, right!
<simpson> Profpatsch: The bad news is that (WLOG) Python packagers upstream are terrible about having clean tarballs.
<simpson> I can see a linter-ish kind of thing being useful, where you explicitly ask for a strict use-all-of-the-tarball mode, but I can't see us comfortably requiring that across all of nixpkgs.
<Sonarpulse> Profpatsch: In general...I'm all for more rigidity!
<gchristensen> is it possible to override the S3 endpoint for an S3-based binary cache in Nix?
<{^_^}> #44107 (by dezgeg, open): Revert "nixos: rename system.{stateVersion,defaultChannel} -> system.nixos.\1"
<{^_^}> #44165 (by LnL7, open): lib: remove deprecation warning for nixpkgsVersion
<LnL> they are not ever really consistent, one adds a prefix and the other removes it
<gchristensen> I'm pro reverting the renames
Lisanna has quit [Ping timeout: 255 seconds]
orivej has quit [Ping timeout: 268 seconds]
<clever> LnL: i noticed that renaming system.stateVersion broke nixops, because the automatic rename modules are not bidirectional
<clever> LnL: i think you can set the old name, but you cant read the old name, and then nixops needs a big hairy if statement to handle both versions of nixpkgs
<gchristensen> let's go back
<clever> keeping the original name seems best in my view
<gchristensen> +1
<gchristensen> Dezgeg: merge! merge! merge!
<clever> and for nixpkgsVersion, i obeyed the deprecation notice, and gave somebody directions based on the new name, where it then failed :P
<gchristensen> :|
<clever> so you must use the old name, or you risk it breaking for old users
<clever> and just put up with the trace from the new code
<gchristensen> yay Dezgeg
<clever> bbl
<Dezgeg> now we just need to document 'renames considered harmful' somewhere
infinisi1 is now known as infinisil
<gchristensen> ,renames=considered harmful
<Dezgeg> does some language or popular (stable) library have some declaration of API stability that we could steal?
Lisanna has joined #nixos-dev
<LnL> rust does have some interesting ideas wrt experimental features
Lisanna has quit [Quit: Lisanna]
biopandemic has joined #nixos-dev
Lisanna has joined #nixos-dev
<LnL> with that model we could, eg. delete stuff that's in master but not considered stable enough for 18.09
<clever> Dezgeg: android for example will rarely (if ever?) modify the arguments to a function, and will only delete functions with plenty of warning, and all functions are clearly labeled with an api version it was introduced in
<clever> Dezgeg: though java has the benefit of allowing method overloading and types to auto-detect which method to call
<Profpatsch> Oh *crap*
<Profpatsch> How can it be that I need to run a f* Makefile to build the nixkpgs docs?
<Profpatsch> That’s a joke, right?
<Profpatsch> I need to go to bed.
<Profpatsch> Ah, hm, it is building with nix-build.
<Profpatsch> I need to go to bed.
sir_guy_carleton has joined #nixos-dev
<gchristensen> Profpatsch: happy to workwith you to make them better.
<samueldr> I may soon have a nixos newbie that will go through the nixos installation process with me watching for stumbling blocks about the documentation
<gchristensen> yay!
<samueldr> (and through nix learning)
goibhniu has quit [Ping timeout: 240 seconds]
the has quit [Ping timeout: 244 seconds]