<niksnut>
the script is supposed to do that automatically...
<qyliss>
I've reverted
<qyliss>
apparently there's a patch that doesn't apply any more too
<dotlambda>
the script should make a PR instead of pushing
<qyliss>
niksnut: did you read my commit message? there's still a patch that doesn't apply
<qyliss>
please do this through a pr
<hexa->
there is `gh pr create` for that ^
<niksnut>
okay, somebody else can update it, I don't really care
<arianvp>
the fact that multiple people stumbled over this gives me a bit of hope that "multiple eyes" principle for supply-chain issues really does seem to work :)
<qyliss>
yeah, this is good
<ma27[m]>
is there somebody working on this? otherwise I can file an update-PR (including a backport PR)
<qyliss>
I'm about to open a PR
<ma27[m]>
qyliss++
<{^_^}>
qyliss's karma got increased to 146
<ma27[m]>
thanks! :)
<arianvp>
where does the update script live? adding a `gh pr create` to that would indeed be neat
<qyliss>
niksnut: thanks for making the release, anyway
<qyliss>
that's gonna help a lot of people out
<hexa->
indeed
<gchristensen>
what's the good stuff in this release?
<qyliss>
fix for Nix#4709 is the big one I'm aware of
<arianvp>
what's the point of the fallback paths if they dont match the paths that get evaluated in nixpkgs? those fallbacks will never be in nix store then
<qyliss>
Yeah that's what's confusing me
<dotlambda>
But if they were the same, wouldn't they have to be changed every time stdenv is?
<qyliss>
I think so
<qyliss>
but the question about how these store paths end up on users' systems remains
<dotlambda>
the installer could download them explicitly
<qyliss>
oh, yeah, I think it does
<qyliss>
but does it make a GC root for them? Wouldn't they get garbage collected?
<qyliss>
or are these only used in the installer?
<puck>
i just checked, and a fallback path is in my nixos-rebuild, but doesn't exist on my system
<LinuxHackerman>
it's obtained by `nix-store -r` in nixos-rebuild if necessary
<LinuxHackerman>
i.e. nix-store -r $fallbackPath → fallback nix is fetched from the cache
<LinuxHackerman>
"interesting", because it's hardcoded to use cache.nixos.org
<qyliss>
how does that help, though, if the purpose is to be a fall back from when Nix isn't in your path like arianvp said?
<puck>
it's to fall back if your nix is too old to evaluate new nixpkgs/whatever
<qyliss>
ohh
<qyliss>
okay
<puck>
i think
<arianvp>
ah so slightly different reason than I said
<qyliss>
reading nixos-rebuild...
<qyliss>
first it tries to evaluate config.nix.package in NixOS, then nix in nixpkgs, then the fallback paths
<V>
Yeah, it's for bootstrapping
<V>
The opacity is unfortunate but it's not insane
<qyliss>
and _then_ it'll fall back to the hardcoded ones
<V>
What would be great would be a simple reproducible script for bootstrapping Nix, like the Guix/bootstrappable people have
<dotlambda>
The fallback paths list aarch64-linux and x86_64-darwin but those seem not to be supported as fallbacks by nixos-install.
<qyliss>
dotlambda: they would be used by L375 of nixos-rebuild.sh though
<qyliss>
but yeah, they're not in prebuiltNix for some reason
m1cr0man has quit [Quit: G'luck]
<qyliss>
oh actually nixSystem assumes you're on Linux, so the Darwin ones still wouldn't be
<qyliss>
which makes sense given this is NixOS
<gchristensen>
I wouldn't be surprised if other tooling also knew to look at the fallbacks
johnny101 has quit [Ping timeout: 260 seconds]
<qyliss>
it does sound like we should teach nixos-rebuild about the fallbacks on aarch64 though
<gchristensen>
yea
<qyliss>
that feels like an oversight
<qyliss>
am I likely to get a Darwin build for my PR, btw?
<gchristensen>
maybe the RFC for architectur esupport should be amended to include this
<qyliss>
gchristensen: well, we could change it now, since the binaries are already generated
<gchristensen>
yes but I mean as criteria for "do we really support it"
<gchristensen>
to make sure it gets done
<qyliss>
oh, yes, good idea
<gchristensen>
that way we don't forget it for power9-linux
<qyliss>
yeah
<qyliss>
although I do wonder if we could make it not use a hardcoded set of architectures
m1cr0man has joined #nixos-dev
<qyliss>
got a Darwin build, yay
<qyliss>
any last words before I merge the Nix update?
<qyliss>
I wonder if the script broke because of name -> pname
<qyliss>
I wonder if there's a GHA that posts comments on commits that have been reverted, with links to the commit that reverted them
<qyliss>
would save gchristensen having to do it :P
<gchristensen>
:D
<sterni>
qyliss: most likely it broke because of that
<sterni>
so I accept all the blame for today's incident :p
<qyliss>
PR to make prebuiltNix support aarch64: #122987
<gchristensen>
there's been a bunch of chat on this in #nixos-chat
<gchristensen>
> yeah, I first learned of this yesterday... my inclination is to stay calm through initial tumultuousness, and unless service really falls apart right away, try to ride it out and take careful steps, useful to note that the fuchs post starts with a header saying it is a draft, and fuchs hasn't resigned ... ideally this channel would stay relaxed about the whole thing until we get past the
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):494:5
<gchristensen>
shitstorm and in to results
<domenkozar[m]>
well we know there's E2E encryption and public internet
<domenkozar[m]>
:D
<gchristensen>
^ that bit I pasted is a combination of a few messages I sent to -chat
<domenkozar[m]>
seems like Matrix isn't showing me history
<{^_^}>
#122866 (by sternenseemann, 20 hours ago, open): llvmPackages_12.lld: make sure llvm libunwind is passed
tomberek has joined #nixos-dev
<s1341_>
sterni cool. I see you're waiting for Ericson2314
<s1341_>
I had also pinged him.
<sterni>
s1341_: I'm not sure we need to wait for them
<s1341_>
Yeah. I'd merge it if you can.
<sterni>
I'm 90% sure this was an accident while making stuff more consistent
<sterni>
lld 12 is the only one which needs libunwind it seems
<s1341_>
yeah. I know Ericson2314 has been working on it a bit making the llvm derivations more consistent...
<s1341_>
obviously someone missed the fact that 12 needs libunwind...
<sterni>
on master it works still
<sterni>
because it was passed originally
<sterni>
I'm gonna wait for the build on ofborg I guess and maybe for a review :)
<qyliss>
if it was passed originally and removal broke it, sounds like readding it should be pretty safe?
ajs124 has quit [Quit: Bridge terminating on SIGTERM]
Scriptkiddi has quit [Quit: Bridge terminating on SIGTERM]
das_j has quit [Quit: Bridge terminating on SIGTERM]
ajs124 has joined #nixos-dev
Scriptkiddi has joined #nixos-dev
das_j has joined #nixos-dev
<s1341_>
i agree.
<sphalerite>
@hydra admins: could I have restart-jobs permissions on my github-linked account? I already have it on my (work) google account, but prefer using the github one now that that integration exists.
<s1341_>
hrm. It looks like i don't have a cross-ar for aarch64-linux...
<s1341_>
it looks like I don't have any binutils at all, although I have a bintutils-wrapper..
<s1341_>
and even a binutils-wrapper-wrapper....
<sterni>
s1341_: are you using lldClang?
<sterni>
a bintools-wrapper-wrapper broke this kind of stuff for me in lldClang (see #122974)
<{^_^}>
nix#4770 (by Ma27, 1 week ago, open): Source complete env in `nix-shell` with `__structuredAttrs = true;`
<s1341_>
sterni: I have to go. I'll chat with you later.
justanotheruser has joined #nixos-dev
<gchristensen>
drakonis has quit [Quit: WeeChat 3.1]
drakonis has joined #nixos-dev
xwvvvvwx has quit [Ping timeout: 268 seconds]
xwvvvvwx has joined #nixos-dev
<abathur>
grumble
xwvvvvwx has quit [Ping timeout: 268 seconds]
xwvvvvwx has joined #nixos-dev
johnny101 has joined #nixos-dev
cole-h has joined #nixos-dev
Jackneill has quit [Remote host closed the connection]
<sterni>
s1341_: nix-build -A pkgsCross.aarch64-android.stdenv.cc.bintools.bintools && ls result/bin/aarch64-unknown-linux-android-ar works as expected
rj has joined #nixos-dev
<sterni>
can't build bionic though
<sterni>
because of a hash mismatch
<sterni>
are you on staging?
Church- is now known as Morpheus
Morpheus is now known as Church-
<regnat[m]>
<ma27[m] "niksnut: Regnat anything else "> Sorry, I’m off for the end of the week so I didn’t have time to review it yet. Will have a look on monday unless Eelco beats me to it
<Mic92>
domenkozar[m]: did you just get the rpi4 to write the tutorial?
<domenkozar[m]>
yeah
<sphalerite>
gchristensen: you're a hydra admin right? :D
<gchristensen>
yeah
<sphalerite>
could I have restart-jobs permissions on my github-linked account? I already have it on my (work) google account, but prefer using the github one now that that integration exists.
* gchristensen
looks at the policy
<gchristensen>
yes
<gchristensen>
PM me your associated email
<jonringer>
just a heads up, last staging-next cycle will begin in a few hours. If you have any staging PRs that you would like to see included, get them merged in the next few hours
<jonringer>
last staging-next cycle before branch-off*
<domenkozar[m]>
jonringer: hmm, I thought that deadline was two weeks ago?
<domenkozar[m]>
thefloweringash: any chance of merging apple silicon work?
Church- is now known as JJJ
orivej has joined #nixos-dev
pmy has quit [Ping timeout: 265 seconds]
pmy has joined #nixos-dev
rajivr has quit [Quit: Connection closed for inactivity]
disasm_ has quit [Quit: WeeChat 2.0]
disasm has joined #nixos-dev
<dotlambda>
domenkozar[m]: The deadline for breaking changes (including the Apple Silicon work I guess) was two weeks ago. This is the last chance for mass-rebuild fixes.
<hexa->
apple silicon wouldn't be breaking, since it it strictly speaking doesn't break anything that worked before
<dotlambda>
It's labeled as rebuilding stdenv though.
<samueldr>
assuming it only introduces things for apple silicon and strictly doesn't change the behaviour otherwise :)
<jonringer>
rebuilding stdenv is fine. My main concern is with regressions
<jonringer>
breaking changes to a package, or restructing a commonly used package will likely cause a lot of work to be done on master to fix all the breakages. That's what I'm trying to avoid :)
<MichaelRaskin>
Also, assuming it works as merged, because otherwise there
<MichaelRaskin>
is a lot of further changes that will be wanted as backports…
<MichaelRaskin>
Even if aarch64-darwin work doesn't technically break anything on x86_64-darwin
JJJ is now known as Church-
cole-h has quit [Ping timeout: 260 seconds]
<rmcgibbo[m]>
i suck. i wanted to get 113220 in before 21.05, but... not going to happen.
<gchristensen>
that is surely not an "i suck" :)
<rmcgibbo[m]>
:) just trying to keep all the balls in the air. sometimes a couple fall.
<gchristensen>
of course
<MichaelRaskin>
_Or_ go orbital and fail to land in time
<{^_^}>
#110826 (by jcleng, 15 weeks ago, open): package php-with-extensions-7.3.25 be without phpize,and no php-dev package
<mjsir911>
${php.dev}/bin/php-config has a hardcoded `php_binary` path to ${php.out}/bin/php, and doesn't re-point to php.withExtensions paths, so the pointed binary doesn't have additional extensions
<mjsir911>
Does it make sense to patch php-config with a PATH based php-binary?
<mjsir911>
I think I understand what's going on, I just don't know the right way to go about with a solution
plumm has quit [Ping timeout: 268 seconds]
plumm has joined #nixos-dev
rj has quit [Ping timeout: 240 seconds]
supersandro2000 has quit [Killed (orwell.freenode.net (Nickname regained by services))]