<domenkozar>
we actually have a retry that parses logs for hercules due to this bug
<domenkozar>
:@
FRidh has joined #nixos-dev
johanot has joined #nixos-dev
Jackneill has quit [Quit: Leaving]
Jackneill has joined #nixos-dev
johanot has quit [Ping timeout: 255 seconds]
johanot has joined #nixos-dev
init_6 has joined #nixos-dev
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-dev
orivej has quit [Ping timeout: 244 seconds]
__Sander__ has joined #nixos-dev
FRidh has quit [Ping timeout: 250 seconds]
copumpkin has joined #nixos-dev
johanot has quit [Quit: WeeChat 2.2]
<gchristensen>
qyliss, sphalerite: if you'd be a dear and call FCP that'd be delightful :)
asymmetric has joined #nixos-dev
<domenkozar>
hmm, shouldn't command-not-found recommend bin output?
asymmetric_ has quit [Ping timeout: 264 seconds]
<gchristensen>
nix-env can't pick outputs iirc
<domenkozar>
$ dig
<domenkozar>
The program ‘dig’ is currently not installed. You can install it by typing:
<domenkozar>
nix-env -iA nixos.bind
<gchristensen>
ugh that one is horrible
<domenkozar>
$ nix-shell -p bind --run "dig"
<domenkozar>
/tmp/nix-shell-31821-0/rc: line 1: dig: command not found
<gchristensen>
-p bind.dnsutil
<gchristensen>
-p bind.dnsutils
<domenkozar>
yeah I know, but it should index what output it belongs to
<domenkozar>
:)
<gchristensen>
or undo that unholy split
<domenkozar>
output indexing seems more universal?
<gchristensen>
only if we have unholy splits! (I agree)
<domenkozar>
it's hard to enforce discipline
<domenkozar>
easier to each computers to be smarter
<domenkozar>
:)
<gchristensen>
hmm true, we could teach stdenv to prevent bin/* in any output other than ot
* gchristensen
will stop
<domenkozar>
I like that
<ekleog>
gchristensen: well, it's basically what nix does: teach computers to enforce discipline
<ekleog>
whether we actually want that is another question
<gchristensen>
of course
<domenkozar>
I wonder how many packages would break though
<gchristensen>
only one way to find out, commit and push to master
<domenkozar>
maybe it's worth it also for sake of nix-env?
<gchristensen>
nah, nix-env needs fixing anyway so people can get manuals
<domenkozar>
I wonder what's the balance between fixing 200 packages to comply
<domenkozar>
or just allow it everywhere and support that
* ekleog
likes multiple-output derivations as precisely split as possible
<domenkozar>
well basically as soon as you have a use case where you can't
<domenkozar>
you break command-not-found
<domenkozar>
now, I don't know why you would wan't to
<domenkozar>
:D
<gchristensen>
heheh
<domenkozar>
one reason would be closure size
<ekleog>
a lot of closure size, but also having exactly the right executables and nothing more (which makes sense in a noexec-with-CLIP's-patch system where root is the only one allowed to talk to nix-daemon)
<gchristensen>
maybe... but it is sort of an important part of how the user experiences Nix. could you imagine having to nix-env --uninstall coreutils-ls coreutils-rm coreutils-touch
<gchristensen>
or worse, nix-env --install coreutils-ls coreutils-rm coreutils-touch
<ekleog>
actually thinking about it again, at times I think I'd like to just be able to install part of a derivation
<ekleog>
heh, I was actually thinking of splitting this way
init_6 has quit [Ping timeout: 250 seconds]
<ekleog>
but we'd have to have a way to not install only `out` by default, but to have the derivation specify what its default set of (un)installed outputs would be
<gchristensen>
you might be able to do the dirty deed on your own with copy and patchelf and binary sed
<gchristensen>
and a gob-smacking quantity of IFD
<ekleog>
weeeeeell… yes, but… :p
<ekleog>
anyway, my thoughts are more longer-term thoughts, as they'd require changes to nix (to be able to install only part of a derivation instead of having the derivation split itself between multiple outputs)