| 9 Jun 2023 |
 infinisil | Yeah sounds good | 19:28:54 |
| infinisil | I'm a bit sad to see unit go, I think it would make a lot of sense going forward, but I agree that something like by-name would be more agreeable | 19:30:55 |
| infinisil | In the end I don't think it matters very much though, and I hate to have to discuss this trivial aspect of the RFC so much, this ate away a number of hours today. | 19:32:24 |
| infinisil | Also something to consider: If we do move to a more unit-like structure in the future (not just packages), we can still perform the rename then if appropriate pkgs/by-name -> unit` | 19:40:05 |
 Robert Hensing (roberth) | Yeah, it doesn't have to be unit for that. Also we can discuss it all over again then :p | 19:41:34 |
| Robert Hensing (roberth) |
I'm a bit sad to see unit go
A love/hate relationship here. It was a nice compromise and a reminder of a larger vision, but ultimately too meaningless it seems. It's been a central common ground almost in the team. Maybe Stockholm syndrome? We did discuss naming for a very long time... | 19:43:55 |
| infinisil | Robert Hensing (roberth): Are you perhaps already writing a github comment to propose the by-name idea? | 19:45:55 |
| Robert Hensing (roberth) | I was thinking to, but then I didn't come up with an actual comment | 19:46:19 |
| Robert Hensing (roberth) | I guess a practical downside of by-name is that a "by-name directory" sounds kind of awful | 19:47:01 |
| Robert Hensing (roberth) | I'd be ok with keeping "unit directory" around as a term just for the RFC | 19:47:18 |
| Robert Hensing (roberth) | I don't think we need to use it much outside the RFC's context | 19:47:28 |
| infinisil | "packages by name" sounds pretty good | 19:47:32 |
| infinisil | Spelled out pkgs/by-name | 19:47:38 |
| infinisil | Hmm, not really a term to use when referring to it though | 19:47:54 |
| Robert Hensing (roberth) | "package directory"? | 19:48:12 |
| infinisil | That's just pkgs, it's a directory! | 19:48:28 |
| Robert Hensing (roberth) | Hmm sounds like that, but not if clearly articulated or written | 19:48:53 |
| infinisil | Anyways, yeah it doesn't really matter how we refer to it, if at all | 19:49:01 |
| infinisil | I started writing the implementation of the first PR and made sure to not use unit directory in the CI error messages | 19:49:28 |
| Robert Hensing (roberth) | Yeah it's not a term people should have to learn | 19:50:05 |
| infinisil | If anybody wants to check out the start of the WIP implementation: https://github.com/tweag/nixpkgs/tree/spp-1, I'll create a PR hopefully next week | 19:50:21 |
| Robert Hensing (roberth) | Just like how "store dir" is irrelevant to most store users | 19:50:30 |
| infinisil | Minor advantage to by-*, it's ordered fairly early, compared to unit which is last :P | 19:52:25 |
| infinisil | Considering that we have a potentially better alternative now and more people think unit isn't great, I think that's reason enough to cancel FCP for now, discuss this to completion, then do another FCP once settled. | 19:54:56 |
| Robert Hensing (roberth) | Oh looks like I wasn't the first to think of it https://github.com/NixOS/rfcs/pull/140#discussion_r1224642417 | 19:57:27 |
| infinisil | By about 30 minutes haha | 19:58:36 |
| infinisil | So easy to miss these threads.. | 19:58:56 |
| infinisil | Plug for https://github.com/NixOS/rfcs/pull/138 | 19:59:02 |
| Robert Hensing (roberth) | In reply to @infinisil:matrix.org
Considering that we have a potentially better alternative now and more people think unit isn't great, I think that's reason enough to cancel FCP for now, discuss this to completion, then do another FCP once settled. I feel like FCP should allow a rename to happen without a restart, but maybe I'm wrong | 19:59:10 |
| Robert Hensing (roberth) | repos are for suckers who aren't telepathic | 19:59:50 |
