<{^_^}>
[patchelf] @rcombs opened pull request #166 → Fixes for PIE executables → https://git.io/fjThY
drakonis has joined #nixos
o1lo01ol1o has joined #nixos
Ariakenom has quit [Quit: Leaving]
drakonis_ has quit [Ping timeout: 240 seconds]
<aanderse_>
aanderse
<aanderse_>
... wrong window :)
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<elvishjerricco>
To which underlying file systems do sandboxed builds write during the build? I assume it's just whatever file system /nix/store is on, so the output directories can be instantly moved from the chroot?
<elvishjerricco>
And does that include the NIX_BUILD_TOP, or does that go to /tmp?
Fare has joined #nixos
goibhniu1 has quit [Ping timeout: 245 seconds]
domogled has quit [Ping timeout: 255 seconds]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fjThV
groodt has joined #nixos
<groodt>
Hi. Is this the right place to ask about Darwin problems on Hydra?
<gchristensen>
sure, though if it requires very specific knowledge you should know the only admin still awake is about to go sleep
<groodt>
Thanks gchristensen. All good if people need to sleep of course. The question is around Darwin builds in sandbox and the jdk. I don't think it works on Hydra. Related to: https://github.com/NixOS/nixpkgs/pull/58557
<{^_^}>
#58557 (by groodt, 4 hours ago, open): bazel: 0.22.0 doesn't build in Darwin sandbox
silver has quit [Read error: Connection reset by peer]
<groodt>
I guess all the Darwin experts have gone to bed? :)
mabel has quit [Changing host]
mabel has joined #nixos
<mabel>
this is a pretty transcontinental bunch, I bet someone will be online later.
<groodt>
thanks!
<groodt>
I just need somebody who can kick off a Hydra build for the moment
<mabel>
personally, I'm in need of an npm expert. I'd like to build a VS Code extension from scratch, rather than pulling from the gallery. there's a nice little toolchain for npm packages in nix, but it represents each dependency as a derivation.
<mabel>
unfortunately, the npm dependencies for even a simple extension number in the high hundreds or low thousands
Supersonic has joined #nixos
<groodt>
mabel, is the issue generating the derivations?
<InFlames>
I did see that and node2nix, and while that's an interesting project, I feel like it breaks the ecosystem is ways that I'm not entirely sure I understand yet
<simpson>
Happens. npm deserves to be broken.
<simpson>
Anyway, that's the recommended way to do things. Or at least the documented way.
<InFlames>
this looks helpful, looks like I missed a few things last night, I'll try again
<InFlames>
thank you
<InFlames>
npm isn't perfect, but asking the ecosystem to bend to nix seems kinda hostile for anyone not on nix
<InFlames>
now, I don't really know if that's required, but it did kinda seem like it was the idea
<InFlames>
as for my own projects, it could be nice
<simpson>
This is the nature of systems like Nix which obey the Principle of Least Authority: Many tools used to an ambient authority will break.
<simpson>
The fact that some things are forbidden is precisely the strength of Nix.
<InFlames>
I don't disagree
<InFlames>
and as long as that solution won't impair use outside of nix, I'm all for it and perhaps should look back into it
<simpson>
Oh, I suppose that we shouldn't confuse Nix's general capability-like design and POLA awareness with my own personal hostility towards npm/etc.
<InFlames>
haha, I don't blame your hostility towards npm, they finally flattened the dependency tree and we still can't get the same package versions on all machines :/
<InFlames>
oh, and added the lock file
<InFlames>
I do wish fb didn't write yarn and instead worked to fix npm, but maybe that's just a choice I don't understand due to a lack of insight
<InFlames>
I am happ to be back on nixos after 2 years away :D
<InFlames>
it's gone a lot better so far, ahah
<InFlames>
I was not ready last time
<simpson>
Heh. I work on stuff like https://monte.readthedocs.io/en/latest/ and so my opinions are a little spicier. For example, require() should be removed entirely from every JS implementation.
<InFlames>
I think I'm just highly confused how this all works no matter what I am reading
<InFlames>
but I'll read more if there is any example on how this works
<InFlames>
I know those 2 node2nix commands don't need to be run together, I ran them as two different attempts and tried them inside and outside the shell
<InFlames>
^ i tried to follow some of that elvishjerricco
slack1256 has quit [Remote host closed the connection]
<InFlames>
like currently I have a devDependency, browserify, and an npm script that calls it to bundle the package, but it says the command is not found
slack1256 has joined #nixos
ricekrispie2 has joined #nixos
ricekrispie has quit [Ping timeout: 245 seconds]
linarcx has joined #nixos
<groodt>
anybody here got the appropriate permissions to kick off a Hydra build?
<linarcx>
Guys. How to find all packages that installed as dependecies of a derivation?
<groodt>
apologies for the spam, anybody available to kick off a Hydra build for me?
drakonis has quit [Ping timeout: 250 seconds]
<CMCDragonkai>
When I used `haskell.lib.buildStrictly` it complains about `error: [-Wunused-top-binds, -Werror=unused-top-binds]`
<CMCDragonkai>
How to allow haskell build to allow unused top binds?
drakonis_ has quit [Ping timeout: 250 seconds]
clr__ has joined #nixos
jbgi has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
init_6 has quit [Ping timeout: 245 seconds]
vk3wtf has quit [Ping timeout: 258 seconds]
clr_ has quit [Ping timeout: 250 seconds]
init_6 has joined #nixos
linarcx has joined #nixos
palo1 has joined #nixos
palo has quit [Ping timeout: 255 seconds]
palo1 is now known as palo
<smolboye[m]>
hello
<smolboye[m]>
does anyone else have this issue where starting firefox before connecting to the wifi causes it to keep showing "problem loading page" even after connecting to the wifi
ninjin has quit [Ping timeout: 256 seconds]
linarcx has quit [Quit: WeeChat 2.4]
linarcx has joined #nixos
endformationage has quit [Ping timeout: 250 seconds]
Erasmus has joined #nixos
linarcx has quit [Ping timeout: 250 seconds]
linarcx has joined #nixos
init_6 has quit []
init_6 has joined #nixos
init_6 has quit [Client Quit]
kleisli has quit [Ping timeout: 272 seconds]
<linarcx>
Anyone here use rust, rls, coc-rls together?
<clever>
that puts the first preempt_schedule_notrace within the c80 bock
schneid3306 has quit [Quit: ZZZzzz…]
<clever>
and looking back at the source, the write and read blocks, simply set some flags, no function calls
<clever>
ioctl will then do some calls and more switch
m0rphism has quit [Quit: WeeChat 2.2]
__monty__ has quit [Quit: leaving]
<clever>
gchristensen: of note, __vdev_disk_physio is static, so the compiler might inline it
m0rphism has joined #nixos
<clever>
gchristensen: i think i would need a debug build to get more info, the compiler can add line number<->addr mappings, that greatly simplify this
mabel has quit [Ping timeout: 245 seconds]
<gchristensen>
nice
<gchristensen>
ok, I'm working to grok what you've said already so I can learn some of this :)
<gchristensen>
is there a trick for getting a debug build of a kernel module? separateDebugInfo = true ?
<monokrome>
seems like xlibs.libXext.dev is broken maybe?
<monokrome>
/nix/store/8fq3isj94jy82ma6axw9hpp10kcjzsjj-libXi-1.7.9-dev/include/X11/extensions/XInput2.h:32:10: fatal error: X11/extensions/Xge.h: No such file or directory
<monokrome>
#include <X11/extensions/Xge.h>
<monokrome>
/nix/store/8fq3isj94jy82ma6axw9hpp10kcjzsjj-libXi-1.7.9-dev/include/X11/extensions/XInput2.h:32:10: fatal error: X11/extensions/Xge.h: No such file or directory
<monokrome>
So, that gets to the same error again? O_o
<monokrome>
seems like it's on the fs in /nix/store/niwyvr5cxsj5mzh37dhzi4zdpdgwqndv-libXext-1.3.3-dev/include/X11/extensions/Xge.h/nix/store/niwyvr5cxsj5mzh37dhzi4zdpdgwqndv-libXext-1.3.3-dev/include/X11/extensions/Xge.h
<monokrome>
eep sorry, I keep double copying >.<
Myhlamaeus has quit [Ping timeout: 240 seconds]
<clever>
monokrome: echo $NIX_CFLAGS_COMPILE
<monokrome>
mm I added that output to the gist
<monokrome>
reading now
<monokrome>
that path isn't in it :o
<clever>
monokrome: what args did you run nix-shell with?
o1lo01ol1o has quit [Remote host closed the connection]
kreisys has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Aleksejs>
Hi, Can someone help me? I'm rebuilding nixos on RPi 3 model B, but new generation does not appear and changes are not available after restart
domogled has joined #nixos
<clever>
Aleksejs: is /boot/ correctly mounted?
<Aleksejs>
clever: I guess. I followed instructions on nixos.org
<clever>
Aleksejs: double-check it in `mount`
<Aleksejs>
clever: /dev/mmcblk0p1 on /boot vfat
o1lo01ol1o has joined #nixos
<Aleksejs>
Also, thebsecond partition is mounted as both / and /nix/store - is it right
<Aleksejs>
?
<gchristensen>
yea
<Aleksejs>
Sorry for typos, I'm on the phone
domogled has quit [Ping timeout: 255 seconds]
domogled has joined #nixos
domogled has quit [Client Quit]
o1lo01ol1o has quit [Ping timeout: 246 seconds]
npmccallum has joined #nixos
linarcx has joined #nixos
<samueldr>
:/ the logo was mirrored in that slide!
rprije has quit [Remote host closed the connection]
rprije has joined #nixos
zupo has joined #nixos
<das_j>
I'm trying to build a static package using `nix-build '<nixpkgs>' pkgsStatic.mumsi`. However, while bootstrapping musl (which is required for the static builds I think), it fails with: attempted static link of dynamic object `/nix/store/w63dw8j8qm3618hfjnsi3fy5lawfnqhz-x86_64-unknown-linux-musl-stage-final-gcc-debug-7.3.0/x86_64-unknown-linux-musl/lib/../lib64/libstdc++.so
<infinisil>
avn: I've been using haskell-ide-engine, which works pretty well with cabal
<das_j>
Oh it also doesn't work with makeStaticBinaries and unstable
<avn>
infinisil: would it works with `stack --nix`?
agander has quit [Quit: Leaving]
<infinisil>
avn: HIE (haskell-ide-engine) also has stack support yes, but I haven't tried it. And I don't like using stack with Nix, doesn't make for a great experience all around
<callmecabman>
has anybody used the nixos+ganeti combo for smallscale virtualization?
goibhniu1 has joined #nixos
<avn>
infinisil: otherwise I will be asked, why I not use any linux well compatible with stack. :/ I prefer personally not use ide at all, but sometimes pressed into it
o1lo01ol1o has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
avn: Ah, NixOS isn't incompatible with stack at all, stack is only a hassle when you want to get nix builds of stack projects
<infinisil>
Afaik
<DigitalKiwi>
avn: dante if you emacs
<infinisil>
avn: HIE is a bit new though, so you might run into some trouble, maybe try something like dante or intero instead (I haven't tried those though)
<infinisil>
And I think VSCode also has pretty good haskell integration with some plugins
<DigitalKiwi>
stack has never worked for me on any os
<avn>
infinisil: I heard a lot from my collegues that I MUST use intero, but when I google how to do it, I haven't see any useable walkthru
<avn>
infinisil: I look on possibility of emacs+doom-emacs+intero if it will works
o1lo01ol1o has joined #nixos
<infinisil>
With lesser-used technologies it is often the case that you won't find an easy walkthrough/tutorial, the official readme's are your best help in such cases
o1lo01ol1o has quit [Client Quit]
<infinisil>
avn: Well it might work, but as with everything in emacs, it might take a bit to set it up how you want. There's also spacemacs which comes with a good default (vim) setup, with haskell support
<infinisil>
And if you don't need to use an IDE(-like thing), you can just go for a simple text editor + ghcid, which i hear is pretty good
<avn>
infinisil: I generally ok with emacs, but both doom/spacemacs is out-of-nix things, and I always worrying how to integrate it. And how feed existing `stack --nix` project to nix-installed intero
mmlb9774508 has joined #nixos
<infinisil>
stack should work out of the box, even on NixOS
<infinisil>
Well, you might need to add a little bit of nix stuff to stack.yaml, but that's very easy
<{^_^}>
[nixpkgs] @Ma27 pushed commit from @dotlambda to release-19.03 « postfix: 3.4.3 -> 3.4.4 »: https://git.io/fjkna
<infinisil>
avn: I guess it would be nice if there was an overview somewhere regarding what Haskell setups work how well with Nix
alex`` has quit [Ping timeout: 246 seconds]
<avn>
infinisil: I currently use `stack build --nix mycode` and it works well (I have some stack-shell.nix bounded, and it works)
Guest2990 has quit [Read error: Connection reset by peer]
sinner has joined #nixos
sinner is now known as Guest58976
Myhlamaeus has joined #nixos
selfsymmetric-mu has left #nixos ["gone to the land of dead hiccups and extinguished light bulbs"]
drvirgilio has quit [Ping timeout: 252 seconds]
hoijui has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Infinisil closed pull request #48012 → redmine: add selectable plugin support at the cost of reproducibility. (FYI only, NOT for merging) → https://git.io/fxsiF
<{^_^}>
[nixpkgs] @worldofpeace opened pull request #58596 → nixos/rename: drop system.nixos.{stateVersion, defaultChannel} → https://git.io/fjkEs
adetokunbo has joined #nixos
<infinisil>
Ralith: Hmm, what's with all the overrideAttrs to change the source? Can't we just use a single version everywhere?
<Ralith>
infinisil: per discusison on the issue, we cannot; each use site of glslang specifically requires a different revision
<Ralith>
blame google's inability to package anything sanely :P
<infinisil>
Ah sorry didn't read properly
<infinisil>
Ralith: That seems good to merge to me then, also good from your side?
biopandemic has quit [Ping timeout: 250 seconds]
<Ralith>
to my knowledge this is as good as the nix idioms are going to get for it, and it's otherwise a mundane update + one significant bugfix
<infinisil>
Oh hmm the headers propagatedBuildInputs thing
<Ralith>
yeah, I'm not sure what the proper idiom is there
<Ralith>
thoughts?
<infinisil>
I'm not sure about that, shouldn't the packages that need the headers include them directly?
<infinisil>
I've never seen propagatedBuildInputs being used like that
<Ralith>
historically the headers were part of the loader package even upstream, so I added that for backwards compat; however, nobody noticed that it was broken in 18.09, so I'm not averse to getting rid of their inclusion entirely
<Ralith>
I don't think it's a terribly weird use of popagatedBuildInputs though
<Ralith>
what's it for, if not that?
wfranzini has quit [Remote host closed the connection]
<infinisil>
Things like python I thought, where if you depend on foo, you also depend on all dependencies of foo, at runtime
<Ralith>
it's build inputs, not runtime dependencies
<infinisil>
Hm yeah
<Ralith>
I think it also comes up whenever you have a C/C++ library whose headers include the headers of some other library, for example
<infinisil>
You mentioned that it fixes a regression in master, what is that?
<Ralith>
vulkan-loader's copy of the headers are currently installed in $out/vulkan instead of $out/include/vulkan
<Ralith>
so clang/gcc won't find them
<infinisil>
I see a $dev output too though
<infinisil>
The headers aren't there either?
<Ralith>
vulkan-loader's copy of the headers are currently installed in $dev/vulkan instead of $dev/include/vulkan, I should say
<Ralith>
or $dev is empty
<infinisil>
Ah
<Ralith>
I forget how it shakes out but it's wrong :P
<Ralith>
they are there, but they are at the wrong path
domogled has quit [Ping timeout: 255 seconds]
<infinisil>
Hmm they're in $dev/vulkan/*.h
<Ralith>
because cp has annoying implicit behavior depending on whether the destination directory exists or not
<Ralith>
yes, which is useless
<infinisil>
Wouldn't a better fix be to just install them in $dev/include/vulkan/*.h?
<Ralith>
why is that better?
<infinisil>
Because that's the standard location, our c compilers and stuff all look for headers there
<Ralith>
they are installed in ${vulkan-headers.dev}/include/vulkan/*.h, which is also the standard location
<Ralith>
using propagatedBuildInputs instead of re-installing them somewhere else removes the opportunity for error
<infinisil>
Huh, so why does vulkan-loaders also have headers? Are these the same headers?
<Ralith>
historically the headers were part of the loader package even upstream, so I added that for backwards compat; however, nobody noticed that it was broken in 18.09, so I'm not averse to getting rid of their inclusion entirely
<Ralith>
that is to say, upstream moved the headers into a new package, and I sought to preserve backwards-compat with the original package
johanot has quit [Read error: Connection reset by peer]
<infinisil>
I see, yeah that sounds better actually
<Ralith>
getting rid of the inclusion entirely?
<infinisil>
Yeah
<infinisil>
People will notice, in which case they can just add `vulkan-headers` themselves
<Ralith>
kk
<infinisil>
We don't need to fix everything upstream messes up ourselves :P
<Ralith>
I spent some time with vulkano but ultimately moved to ash
<samueldr>
reviewing a specific PR, and the packaged font doesn't have a license; I'm unsure whether meta.license should be an empty list or just dropped in that case
jbgi has quit [Ping timeout: 246 seconds]
<Ralith>
vulkano manages to be a thick heavyweight abstraction that requires massive amounts of effort to go off the beaten path with, without actually delivering on its promised safety improvements
<Ralith>
plus with ash everything's 1:1 so you can easily reference standard vulkan documentation and examples
<infinisil>
Ralith: -> #nixos-chat
<{^_^}>
[nixpkgs] @pacien opened pull request #58597 → riot-desktop: use own yarn2nix → https://git.io/fjkEb
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « libextractor: 1.8 -> 1.9 (#58293) »: https://git.io/fjkEN
<rycee>
samueldr: If no license is specified anywhere then I guess it should be treated as unfree. I suspect it technically actually is completely unredistributable.
Fare has joined #nixos
<samueldr>
yeah, that's a given; though does nixpkgs+hydra treat a lack of license the same as unfree+unredistributable?
<samueldr>
if so, when it's "unknown" rather than "unfree", would it be semantically appropriate to not put meta.license, or an empty list?
<samueldr>
just like saying "yes, we don't know the license" instead of "this is unfree"
<rycee>
That I don't know, but it is certainly an interesting question.
<samueldr>
(the nixpkgs manual has nothing specifically about that subject AFAICT)
<samueldr>
(and here they author of the PR I suspect has used "unlicense" thinking it meant "unlicensed" or "unknown", and wasn't the name of a specific license)
<samueldr>
heh, that's something I was going to suggest about how to find more details about that, but also, copyright != license
<samueldr>
and yeah, scict.ir is giving me 403 Forbidden here
<rycee>
I was mainly thinking about the "All Rights Reserved", to me that seems very non-free. From what I can tell it shouldn't be on github at all.
<samueldr>
same
adetokunbo has joined #nixos
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<samueldr>
there is another repo, older, which seems to have the same fonts, as webfonts, but I'm not convinced the license is right https://github.com/kyte-group/persian-fonts
<rycee>
averell: But it just says "All rights reserved", there is no addition that qualifies this statement. The MIT license example is essentially "All rights reserved except for this and this and this…"
<averell>
Yes, but you don't have to say "All rights reserved" to reserve any rights anyway.
<averell>
So if we have other unknown licenses not saying this, that should IMO be treated the same.
<rycee>
Yeah, so it doesn't really add anything :-)
<{^_^}>
[nixpkgs] @xeji pushed commit from @rembo10 to master « sabnzbd: 2.3.7 -> 2.3.8 (#58357) »: https://git.io/fjkur
<averell>
could ping the github guy to add a license to his repo. he was active in january.
<samueldr>
and looks like the scict.ir domain expired and was bought by one of those spam site around 2017-18, which explains the lack of fonts details
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « airsonic: 10.1.2 -> 10.2.1 (#57702) »: https://git.io/fjku9
<samueldr>
yeah, and from the sleuthing I did in check-meta.nix, it looks like that as long as it doesn't have "free = false" anythin in the license field is seen as free
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « bacula: 9.4.1 -> 9.4.2 (#58007) »: https://git.io/fjkud
<samueldr>
qyliss: pretty sure it isn't :/
<rycee>
Probably would be nice to have an entry like `unknown = { fullName = "License terms unknown"; free = false; };` in licenses.
<samueldr>
yeah
<rycee>
And replace the "unknown" with it.
<samueldr>
though `hasDeniedUnfreeLicense` is a bit more complex, and uses that. The lack of of license (rather than "unknown") would be treated as unfree