!djTaTBQyWEPRQxrPTb:nixos.org

Nixpkgs Architecture Team

231 Members
https://github.com/nixpkgs-architecture, weekly public meetings on Wednesday 15:00-16:00 UTC at https://meet.jit.si/nixpkgs-architecture52 Servers

Load older messages


SenderMessageTime
20 Mar 2023
@qyliss:fairydust.spaceAlyssa RossIt does exist, but it's not the primary communication method for the team 20:30:45
@qyliss:fairydust.spaceAlyssa Rossfor example, I follow this channel, but the first I heard that the team was going to be talking about the Rust thing was after the NACK comment was made, on behalf of the team20:31:25
@infinisil:matrix.orginfinisilHmm yeah, that could definitely be done better20:32:19
@qyliss:fairydust.spaceAlyssa RossAnd again, I'm not saying the team shouldn't work this way, that's fine. It's a fine way to get a lot of work done, as can be seem from the impressive results so far. I'm saying that the NAT should not be something that speaks with a collective voice, at least on other people's ideas.20:34:32
@infinisil:matrix.orginfinisil
In reply to @qyliss:fairydust.space
It might be worth discussing though, what's the scope of the architecture team? The web page talks about it as being a group of people who work together to propose solutions to big problems in Nixpkgs. Commenting on other people's proposed solutions as a team, rather than as individual team members with their own opinions, feels not so in line with that, and more like an attempt to establish the team as an authority over Nixpkgs architecture.
I haven't answered this, here's my opinion: I do think the Nix ecosystem at large is lacking a vision of where it wants to go. And I think this is why Nixpkgs is such a mess today, it's the accumulation of a decade of various people pulling it in various directions. The primary goal of this team is to reconcile that, to try to make it not a mess anymore. But a secondary goal for me is to actually give Nixpkgs a vision for the future. And yes, having a vision means being opinionated and actually pulling off that vision requires having some authority over Nixpkgs. My hope is that the community is decently represented in this team so that we can be trusted with actually doing reasonable things.
20:38:33
@qyliss:fairydust.spaceAlyssa Rossbut what gives the NAT the legitimacy to be the group who gets to set the vision, and have that authority?20:39:35
@qyliss:fairydust.spaceAlyssa Rossfrom my perspective, we have produced the most powerful package repository ever, leagues ahead of anything else, precisely because there is no single vision that has to be adhered to20:40:43
@infinisil:matrix.orginfinisilI mean, it's not like a black-white thing, you can either have a vision or not. I think Nixpkgs could benefit greatly from having some things be a bit more guided.20:41:30
@infinisil:matrix.orginfinisil
In reply to @qyliss:fairydust.space
but what gives the NAT the legitimacy to be the group who gets to set the vision, and have that authority?
Inherently nothing, the NAT was bootstrapped from not much. There's no special permissions or anything that it has over others in Nixpkgs. The thing it does have is exactly what the team is doing: Transparency in its actions, accountability, the results we produced, the motivation that drives us, the sponsoring allowing us to work on the team, etc.
20:44:25
@infinisil:matrix.orginfinisil But I'm always open to suggestions, I'm thankful for the feedback regarding the GitHub comment. I think it's okay to not use "we" there 20:47:04
@qyliss:fairydust.spaceAlyssa RossAt this point in the team's existence, it speaking with a collective voice on other people's ideas comes across as an attempt to project authority without having earned it yet. Maybe it would be different if the NAT had an established record of making big, successful improvements that wouldn't have been possible without it, but at this stage the NAT is so new you're still working on the first one. It would be different if NAT members were just speaking as themselves, rather than representatives of the team, of course.20:47:19
@raitobezarius:matrix.orgraitobezarius
In reply to @infinisil:matrix.org
I haven't answered this, here's my opinion: I do think the Nix ecosystem at large is lacking a vision of where it wants to go. And I think this is why Nixpkgs is such a mess today, it's the accumulation of a decade of various people pulling it in various directions. The primary goal of this team is to reconcile that, to try to make it not a mess anymore. But a secondary goal for me is to actually give Nixpkgs a vision for the future. And yes, having a vision means being opinionated and actually pulling off that vision requires having some authority over Nixpkgs. My hope is that the community is decently represented in this team so that we can be trusted with actually doing reasonable things.
I'm not convinced that achieving these goals requires authority over Nixpkgs; it is also possible to drive changes by taking the time to convince the stakeholders.
At least, there should be compelling arguments on why is this necessary to hand over "authority over Nixpkgs" and what do we lose by doing this vs. the current model and what do we obtain with this model.
20:48:07
@qyliss:fairydust.spaceAlyssa RossWe didn't say we wanted to do it as "The Rust Team", even though it could have been done that way. Instead, the proponents all just spoke as themselves, and I think that's more conducive to an inclusive discussion. I don't want to get into a Teams arms race where we form explicit factions with collective opinions.20:49:32
@infinisil:matrix.orginfinisil
In reply to @raitobezarius:matrix.org
I'm not convinced that achieving these goals requires authority over Nixpkgs; it is also possible to drive changes by taking the time to convince the stakeholders.
At least, there should be compelling arguments on why is this necessary to hand over "authority over Nixpkgs" and what do we lose by doing this vs. the current model and what do we obtain with this model.
Well it's only the architecture this team is about, it's not involved in e.g. how packages/modules should be added/updated. It doesn't make sense for it to have authority over the entirety of Nixpkgs. I think it should be more of a "if you want to make changes to these specific files, somebody from the NAT should review it", I guess kind of like code owners
20:55:58
@qyliss:fairydust.spaceAlyssa RossThat sounds a lot more reasonable, but it wouldn't include e.g. the Rust proposal20:56:40
@qyliss:fairydust.spaceAlyssa RossI think it's fine for the NAT to have a more authoratative voice abotu stuff it was responsible for or heavily involved in.20:56:58
@raitobezarius:matrix.orgraitobezarius
In reply to @qyliss:fairydust.space
That sounds a lot more reasonable, but it wouldn't include e.g. the Rust proposal
Well technically increasing significantly the size of nixpkgs could be considered as architecture of nixpkgs
20:57:03
@raitobezarius:matrix.orgraitobezariusBut then, I have concerns about things like splitting nixpkgs, etc. Also, I would ask to clarify how the NAT team would work with the existing RFC process for example.20:57:23
@infinisil:matrix.orginfinisilI think this is something all teams and groups in the community should have, if you're the person knowing about haskell packaging, you should be requested for a review when that part changes. This is kind of a problem with the docs team right now 😅20:57:30
@qyliss:fairydust.spaceAlyssa Rossinfinisil: yeah exactly20:57:37
@qyliss:fairydust.spaceAlyssa Rossraitobezarius: well so far the NAT has just been working on RFCs to propose, which is great!20:58:06
@raitobezarius:matrix.orgraitobezariusSo my question is about: is NAT going to have a 2-level consensus mechanism for large scale changes to nixpkgs? (which are probably any big (re)work on nixpkgs)20:58:59
@raitobezarius:matrix.orgraitobezariusinternal consensus to NAT team → RFC consensus20:59:09
@infinisil:matrix.orginfinisil
In reply to @qyliss:fairydust.space
That sounds a lot more reasonable, but it wouldn't include e.g. the Rust proposal
I think that's in scope for this team because it's about the overall packaging approach, which can have big implications for the future. What if every package manager with a lockfile wants to commit the lockfiles? Would we blow up Nixpkgs too much? Couldn't we have a better sustainable solution for lockfiles? Etc.
21:00:19
@qyliss:fairydust.spaceAlyssa RossYes, it might be in scope for the team, but the team hasn't yet earned any authority over it.21:00:57
@infinisil:matrix.orginfinisilYeah21:01:40
@qyliss:fairydust.spaceAlyssa Ross Maybe I can summarize my concerns. Speaking as a collective team sounds like an attempt to speak authoratatively, almost by definition. Speaking authoratatively in areas where that authority hasn't been earned is going to rub people the wrong way. The Rust thing was an example of that, because the NAT, which has so far (due to being very young, not because you're not doing good work) not done anything in any way related to it. 21:01:53
@infinisil:matrix.orginfinisilIn any case, want me to change the comment? I don't want to make it seem like the team has authority over that, I really just meant the comment as a "we discussed this together and this is what came out of it"21:02:34
@qyliss:fairydust.spaceAlyssa RossNah21:03:03
@qyliss:fairydust.spaceAlyssa RossI just wanted to give you the feedback for future21:03:32

Show newer messages


Back to Room ListRoom Version: 9