gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
zybell_ has joined #nixos-chat
zybell_ has quit [Ping timeout: 248 seconds]
zybell_ has joined #nixos-chat
zybell_ has quit [Ping timeout: 264 seconds]
ma27 has joined #nixos-chat
ma27 has quit [Ping timeout: 256 seconds]
ma27 has joined #nixos-chat
MichaelRaskin has joined #nixos-chat
ma27 has quit [Ping timeout: 256 seconds]
ma27 has joined #nixos-chat
ma27 has quit [Ping timeout: 240 seconds]
jtojnar has quit [Remote host closed the connection]
zybell_ has joined #nixos-chat
ma27 has joined #nixos-chat
ma27 has quit [Ping timeout: 240 seconds]
ma27 has joined #nixos-chat
ma27 has quit [Ping timeout: 255 seconds]
ma27 has joined #nixos-chat
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
infinisil has joined #nixos-chat
<infinisil> Great, ssh port closed on my server after update :/
<infinisil> I could log into digitalocean and start the web console to get it back, but that needs 2 factor authentication via SMS
<infinisil> And I currently don't have any signal :/
<zybell_> for that I had a port open in inetd that restarted sshd simply by connecting to it.
ma27 has quit [Ping timeout: 255 seconds]
ma27 has joined #nixos-chat
ma27 has quit [Ping timeout: 265 seconds]
ma27 has joined #nixos-chat
<infinisil> zybell_: Well I'm testing some changes with sshd_config, sooo it might not be able to restart with that
ma27 has quit [Ping timeout: 246 seconds]
ma27 has joined #nixos-chat
<MichaelRaskin> That's why you add a second sshd when you change the config, and update only one at a time…
zybell_ has quit [Ping timeout: 276 seconds]
<gchristensen> open an SSH connection 6first
<gchristensen> when sshd updates it keeps your ssh connection open, no?
ma27 has quit [Remote host closed the connection]
ma27 has joined #nixos-chat
<MichaelRaskin> gchristensen: that assumes you will never need to reboot
<gchristensen> who reboots?
<MichaelRaskin> People who end up with unkillable processes in D state also holding an important system-wide lock
<gchristensen> (I kid, of coursee)
<gchristensen> I reboot all the time. in a deployment environment, I consider long uptimes a badge of shame
infinisil has quit [Quit: Configuring ZNC, sorry for the join/quits!]
infinisil has joined #nixos-chat
infinisil has quit [Client Quit]
infinisil has joined #nixos-chat
infinisil has quit [Client Quit]
infinisil has joined #nixos-chat
infinisil has quit [Remote host closed the connection]
zybell_ has joined #nixos-chat
infinisil has joined #nixos-chat
infinisil has quit [Client Quit]
infinisil has joined #nixos-chat
drakonis has joined #nixos-chat
taktoa has joined #nixos-chat
Lisanna has joined #nixos-chat
drakonis has quit [Read error: Connection reset by peer]
ma27 has quit [Ping timeout: 276 seconds]
obadz- has joined #nixos-chat
obadz has quit [Ping timeout: 240 seconds]
obadz- is now known as obadz
Sonarpulse has quit [Ping timeout: 264 seconds]
Sonarpulse has joined #nixos-chat
<infinisil> Rebooting has fixed a surprising amount of issues for me..
Drakonis[m] has joined #nixos-chat
<Drakonis[m]> o nice
<Drakonis[m]> smile, you're still an inspiration for red hat :v
<Drakonis[m]> nixpkgs got brought up as an example of a good packaging model
<infinisil> Drakonis[m]: Heh nice
<infinisil> Probably the only package manager that doesn't have proper versioning though haha
<Drakonis[m]> this is the ostree guy
<Drakonis[m]> heh well, being easy to work with through pull requests
<Drakonis[m]> fedora is too feudal
<Drakonis[m]> same with debian
<MichaelRaskin> Frankly, it is a sobering and a stressful reminder that even with all our process bottlenecks…
<Drakonis[m]> debian has remedied the issue by allowing packagers to send in packages using pull requests
<Drakonis[m]> fedora doesn't have that
drakonis has joined #nixos-chat
<drakonis> despite the fact that i'm not a fan of how rigid the system in many aspects, it is still ahead of the curve
<gchristensen> drakonis: where were we brought up?
<drakonis> PRs-on-distgit
<gchristensen> ah cool
<drakonis> it is the direct mention
<drakonis> the other parts are still relevant in a way or another
<gchristensen> having individual repos would make it easier to pretend we had fewer bugs
<drakonis> ha
zybell_ has quit [Ping timeout: 240 seconds]
<drakonis> fedora's packaging stack is quite crummy while debian has a quite robust stack with extremely rigid packaging rules
<drakonis> be back in a few minutes
<drakonis> by virtue of having this repository model, its way easier to maintain packages
<drakonis> because fish was out of date for two years on fedora
<drakonis> only because nobody wanted to go through the process of getting someone to update it
<drakonis> gotta contact the maintainer or go through the red tape involving taking over a package
<gchristensen> yeah
<drakonis> debian has the same issues
<LnL> gchristensen: lol
<MichaelRaskin> That red tape we keep talking about introducing?
<drakonis> except its much easier to do it on your own
<drakonis> what kind of red tape though
<drakonis> debian has the most red tape because you have to write down every possible license a package has
<MichaelRaskin> Well, stronger code ownership model
<gchristensen> that is a good thing IMO
<drakonis> AND you have to package every dependency in the chain
<gchristensen> re licenses
<drakonis> code ownershp model?
<drakonis> yes it is
<drakonis> but there's complex packages with hundreds of varying licenses lol
<gchristensen> yeah
<drakonis> eclipse for example
* gchristensen hides
<drakonis> the debian licenses package is huge
<drakonis> the licenses file
<MichaelRaskin> Well, weakness of the code ownership model is why any PRs ever get merged in Nixpkgs.
<drakonis> the feudalism in fedora and debian is a pretty shitty thing
<MichaelRaskin> I am not sure strong ownership model and feudalism are distinct enough
<drakonis> its a fiefdom
<drakonis> you have hundreds of individual packages with their own maintainers
<drakonis> this is a problem
<MichaelRaskin> Right, clear ownership
<MichaelRaskin> Which is a problem, I agree
<drakonis> people maintaining packages that only have to be version bumped every x time
<drakonis> they tend to create little fiefdoms for themselves
<drakonis> adding red tape is only useful for complex packages sets
<drakonis> because these don't always involve just version bumps when updating
<drakonis> anyways, is this the right place for this conversation or is #nixos-dev the proper place?
<MichaelRaskin> Well…
<drakonis> seems like i already found the answer :V
<MichaelRaskin> Oh?
<drakonis> for my own question
<drakonis> its probably -dev
<MichaelRaskin> I would say that the current relevant topic for process optimisation is the Nix Release Process RFC
<MichaelRaskin> For example: it is the first RFC in half a year that has good chances to be passed after actual comments and significant changes based on discussion
<MichaelRaskin> Yes and no
<MichaelRaskin> I mean, there are some further updates in the other branch, and some plans for even further updates
<drakonis> i assume i am missing something out here
<drakonis> have a link?
<MichaelRaskin> Well, there is a PR against the file you reference in the same repository
<drakonis> ah i see
<MichaelRaskin> There is some discussion in the main repository.
<MichaelRaskin> Yes
<drakonis> this isn't exactly the topic i was talking about though
<drakonis> this is Nix and not Nixpkgs
<MichaelRaskin> Well, having a visibly working RFC process is probably a prerequisite for any process changes anyway
<drakonis> well i see
<drakonis> i don't see any rfc for package ownership :v
<drakonis> i do assume that there's code owners list on nixpkgs
<MichaelRaskin> Also, Nix has the Nix Core team (which is officially declared as a process experiment), so I think it is a good idea to see how it works.
<shlevy> Haven't followed this whole thread, but I plan to have an update against the main PR on Monday
<MichaelRaskin> There is something. It is not clear what it does. The expectation is that _if_ the person is in the list _then_ this person is (one of) the main developers of something complicated
<Drakonis[m]> it looks fine as it is, as long the same courtesy isn't extended to everyone that wants an small package
<MichaelRaskin> It is not clear what ownership means in practice.
<MichaelRaskin> Like, there are notifications, OK
<Drakonis[m]> who maintains what
<MichaelRaskin> We _also_ have meta.maintainers for that
<MichaelRaskin> For complicated packages there is normally no huge competition of people trying to take over that mess…
<Drakonis[m]> it doesn't provide power
drakonis has quit [Read error: Connection reset by peer]
<Drakonis[m]> it simply messages them
<MichaelRaskin> Well, it is probably slightly more rude not to wait a reasonable time for comments by a code owner than by the maintainer
<Drakonis[m]> okay, my point is, don't give people free control over a tiny slice of the repository because they're going to get into fights
<samueldr> hey now, I'll fight your for my right to get into fights over little things (/s)
<Drakonis[m]> this week an unresponsive maintener got in a fight because the folks that took over did it in a way he didn't like and without his consent
<Drakonis[m]> this was debian mind you
<Drakonis[m]> fedora has its own fights too
<Drakonis[m]> drove people out of the project
<MichaelRaskin> It's not like we _never_ have questions «why in the world did you do that this way» answered with «well, we tagged you two weeks ago and got no reply»…
<MichaelRaskin> But that usually doesn't escalate to a proper fight.
<Drakonis[m]> of course
<MichaelRaskin> Our process has many bottlenecks, things get blocked and stalled. Fights are things.
<Drakonis[m]> it's less perceptible
<MichaelRaskin> What? A fight that got stalled before happenning? Yes, it's not always easy to notice.
<Drakonis[m]> the bottlenecks
<gchristensen> `curl https://nixos.org/nix/install | sh -s -- --daemon` :)
<Drakonis[m]> heh
<Drakonis[m]> hmm, how come much of the tooling brought up on nixos weekly isn't available on nixpkgs?
<Drakonis[m]> Profpatsch: look up podman/libpod for the container workgroup
<Drakonis[m]> it's a docker replacement that makes use of k8s pods
<Drakonis[m]> it's by the same people doing cri-o
<Drakonis[m]> also project atomic
<Drakonis[m]> having workgroups is a good move
<gchristensen> re this install question, these problemsare so exhausting
zybell_ has joined #nixos-chat
<Profpatsch> Drakonis[m]: First someone needs to replace our dockerTools image generation with something generating OCI images
<Profpatsch> Until then we can’t really use the fancy OCI stuff
<Profpatsch> I might give that a shot once our project is running nicely with the existing docker images
<gchristensen> is it not OCI?
ma27 has joined #nixos-chat
<Profpatsch> I have a small wrapper here that does stuff like separate contents into its own layers and set up a zombie reaper
<Profpatsch> gchristensen: Nope, it’s the legacy docker image format.
<gchristensen> ah
Sonarpulse has quit [Ping timeout: 256 seconds]
Lisanna has quit [Remote host closed the connection]