Blackraider has quit [Remote host closed the connection]
<abathur>
meh, I'm a bit down today; rough day already and for some reason my source <(curl <nix-bootstrap-script>) is failing specifically on travis-ci macOS jobs but running fine on linux ones
<abathur>
it's been doing it since yesterday, but I was hoping it was just something transient that would work itself out
<bqv>
colemickens: I've tried to compile their develop branch a few times recently, no luck... and I saw the milestone post, but not sure what theyre referring to if anything besides that branch
kroh has joined #nixos-chat
koh has quit [Ping timeout: 260 seconds]
<ldlework>
I'm on a mission to get many more languages working!
<iqubic>
ldlework: What is that you are working on?
<ldlework>
iqubic: Get as many languages I can working in a single org-mode file, via babel.
<iqubic>
Ah. I see. That's a great ambition in life
<ldlework>
I now have awk, bash, C, C++, C#, D, elisp, fortran, javascript, lua, ocaml, perl, python, R, ruby and scheme
<ldlework>
manveru: I'm calling it The Tower of Babix.
xd1le has joined #nixos-chat
malook has joined #nixos-chat
<manveru>
god, terraform is a fickle beast
<ashkitten>
someone tell me why nixops is better than ansible in a practical sense
<Mic92>
you mean using ansible to deploy nixos?
<immae>
ansible is not declarative? if you’re on a debian and you apt-get install a program, then ansible will never care about this new program
<Mic92>
I think it has some declarative resources, but when you remove a package you have to do it explicit, while on NixOS you can just remove the package from your list
<ashkitten>
hmm
maxdevjs has quit [Ping timeout: 272 seconds]
<immae>
That’s what I meant with "not declarative": if you remove some resource from your ansible tasks (with some exceptions), then ansible will *stop caring* about that resource, it will not *actively remove* that resource (unless you ask ansible to actively remove it)
<Valodim>
when ansible fails in the middle of a script, you better hope you end up in a stable state
<Mic92>
It might be interesting though to have an ansible module that can do nixos-rebuild. Ansible runs without nix, which makes it easier to use in a diverse team.
<ashkitten>
i think the person i'm arguing with would just say that these are issues with the ansible scripts not ansible itself
<Valodim>
yes just like segfaults are issues of programs, not C itself
<ashkitten>
she doesn't like rust either...
<Valodim>
yes, well..
<ashkitten>
:(
<immae>
ansible is not "bad", your friend may have other expectations than you. But the main difference is that you cannot leave things behind when you change some config in nix, while with ansible you only care about what you declare in it (and the rest may or may not exist at will)
<ashkitten>
she also says she likes netbsd's reproducible builds better than nix, because it's a traditional package manager
<Mic92>
well ansible cannot be better because it has to fight against the system state of your Linux package manager.
<ashkitten>
and that pkgsrc can do a lot of the same things as nix, including installing multiple versions of the same package
<ashkitten>
it seems neat i guess
<Mic92>
Never used pkgsrc, does it updates atomically?
<ashkitten>
i doubt it
<ashkitten>
it's a classic make install thing
<Mic92>
so it would break the system if pkgsrc fails while installing the new libc?
<ashkitten>
i don't think that's a problem on bsds
<Mic92>
because you never upgrade?
<ashkitten>
because the system is separate from packages
<ashkitten>
and has a separate upgrade process
<Mic92>
that means downtime whenever system security updates are applied?
<ashkitten>
i'm not a bsd expert so i can't comment on that
<Mic92>
Reminds a bit of windows
<ashkitten>
that seems like something that would be a solved problem though
<Mic92>
maybe not as bad though
<ashkitten>
i don't think netbsd is comparable to windows
<Mic92>
I mean the fact that you have to reboot for updates
<ashkitten>
you have to reboot for updates on linux
<ashkitten>
if you update the kernel or critical programs
<Mic92>
For kernel yes, for userland not really
<Mic92>
Also some enterprise distros hotpatch the linux kernel
<ashkitten>
again idk what the process is like on bsds
<ashkitten>
so don't extrapolate from my words
<Valodim>
to be fair, I think nix is definitely still in an early adopter stage. it doesn't really cater to people who aren't interested in shiny and new things
__monty__ has joined #nixos-chat
<ldlework>
oh huh, I just realized I can actually package babix..
<__monty__>
An arduino kernel?
parsley936 has joined #nixos-chat
<joepie91>
IMO morph is a lot closer to a "just works, easy to get started with" tool than NixOps though
<joepie91>
modulo the required knowledge of Nix itself
malook has quit [Quit: malook]
<ldlework>
Can anyone help me think about how to make a derivation that essentially pops out a script that wrap emacs such that it has a number of binaries on its path?
<Valodim>
joepie91: interesting. I'll check that out
<__monty__>
ldlework: I'd grep nixpkgs for wrap. Tons of packages do this.
<__monty__>
gchristensen: Can't you just fork rather than use a subshell?
<gchristensen>
no, because I need to take several steps
<gchristensen>
but it does fork
<__monty__>
How does forking prevent several steps?
<gchristensen>
maybe I don't understand what you're saying
<__monty__>
I guess I'm wondering why it needs to be a subshell and can't just be a fork of the shell, that way you get the child pid.
<gchristensen>
maybe you could write out what you mean{
<gchristensen>
? (dang keyboard swapped { and ?)
<__monty__>
Unfortunately my shellfu does not suffice without sitting down and working through the problem.
endformationage has joined #nixos-chat
waleee-cl has joined #nixos-chat
<colemickens>
bqv: I can't even get the basics of ipfs to work. On my setup, the ipfs config file is owned by ipfs:ipfs and isn't readable by the ipfs group -_-
<bqv>
I have a hacky fix for that
<bqv>
I dunno why it happens
<bqv>
I also don't really want to build another PR
<eyJhb>
bqv: Ohhhh PRs are fun!
<bqv>
No, they really aren't
<eyJhb>
Also, do you have time for a nixus question bqv ? :p Do you deploy using Nixus to your local machine?
<bqv>
colemickens: chmod a+r the config
<bqv>
Yeah, I do
<colemickens>
I could...
<bqv>
rc.fron.io/profiles/networking
<eyJhb>
How do you allow for Nixus to deploy to your local machine? Have you set a authorized keys for root, or just allow for your main user to sudo without any PW?
<eyJhb>
I am a little confused by that
<colemickens>
idk what is going on but everytime I try to touch ipfs stuff hangs
<bqv>
It ssh's to itself, as me, and then does sudo
<colemickens>
after chmod, `echo {} | ipfs dag put` -> hangs
<bqv>
Have you by any chance been deleting pins
<eyJhb>
But it needs to be a passwordless sudo? Right? Do you have your configs online?
<colemickens>
bqv: there's never been any ipfs data on this system
<bqv>
See above
<bqv>
Doesn't mean no pins
<bqv>
The MFS root would have been pinned
<colemickens>
oh ok. I'll check out your ipfs.nix real quick
<bqv>
Deleting that causes a bug
<bqv>
That makes all ipfs commands hang
cole-h has joined #nixos-chat
<abathur>
huh, just got an email from actions about a job that failed 8 days ago
<abathur>
gave me an annoying ~5 minute freakout about how the hell the code was in that state :)
<colemickens>
bqv: sorry if I'm being daft but what is the recovery process if I've gotten myself in this case? I don't see anything in here that looks like an automated fix for "MFS root" (I do see the to/fromipfs scripts, the wrapper, the execstartpost to fixup the config file, etc).
<bqv>
colemickens: the fix is to find the "mfs-replace-root" script on github, and use that to set something that's actually in your repo
<bqv>
The empty directory probably works best
<bqv>
Theres a tracking issue on go-ipfs for this
<colemickens>
:| this is quickly destroying any interest I had.
<colemickens>
I literally just activated this yesterday and have done nothing and it's broken?
<bqv>
I don't think the bug shows up unless you get triggerhappy with deleting pins
<bqv>
Mine's fine now I've stopped doing that, at least :p
<colemickens>
hm something was just jammed in systemd startup
<colemickens>
I stopped one of the sockets and then all of the sudden it finished booting up and put dag put worked???
<bqv>
Heh
* colemickens
its too early in the day for this kind of stuff
<bqv>
Relatable
<bqv>
Sup, lucy
<cole-h>
abathur: I think something whacky's going on with GH's mail
<cole-h>
I just got an email for #97023 (which I manually subscribed to after it was posted) and just now got an email for it as if I had been mentioned in the PR body