| 22 Apr 2025 |
Tristan Ross | Graham shared this https://x.com/grhmc/status/1914679840843719157 | 16:39:51 |
Tristan Ross | So it looks like aarch64 is significantly higher than x86_64 Darwin at least for DetSys | 16:40:31 |
Tristan Ross | Ofc that's not the exact information we need but gives an idea | 16:40:47 |
Tristan Ross | Oh yeah, @[John Ericson]. I know you were involved in the RFC for the platform tier support. I'm wanting to expand upon it and work on a new platform tier list based on this: https://pad.lassul.us/9yYTVp73QBul7dCm6tCWsw?view. I'd appreciate you're feedback and seeing what we can do to improve it. | 17:08:52 |
Tristan Ross | I'm thinking something which lists each tiers and has a breakdown of the support should be clear enough. | 17:09:28 |
John Ericson | I am not at all sentimental about that RFC | 17:10:35 |
John Ericson | revisions are fine! | 17:10:38 |
emily | I do think that any revision should take into account use on the NixOS infra as a criterion | 17:20:23 |
emily | since as long as we need to run aarch64-darwin machines to build anything, there will be expectations relating to it regardless of formal tier | 17:20:51 |
Tristan Ross | In reply to @emilazy:matrix.org I do think that any revision should take into account use on the NixOS infra as a criterion Yes, that's kinda why Ofborg + Hydra is mentioned | 17:24:16 |
Tristan Ross | Having Ofborg support is kinda like a baseline for Hydra. It means we do have machines and some sort of CI. | 17:24:51 |
Alyssa Ross | I kind of feel that to be in whatever the top tier is it should be possible for contributors to reproduce issues on that platform — in a VM or cross is fine, but having to obtain a license for a proprietary OS is not. | 17:29:01 |
Alyssa Ross | Community builder mitigates this to an extent but isn't open to everyone and I'm not sure if it also comes with EULA obligations on its users. | 17:29:35 |
Alyssa Ross | Like sometimes it's just not really fair to expect contributors to do things with Darwin in the way we can for Linux. | 17:30:14 |
emily | I don't think that would be the case if half of all users and contributors were on Darwin though | 17:30:46 |
Alyssa Ross | The fact that the platform can break underneath us is also a concern for calling things "Tier 1", IMO. | 17:30:51 |
Tristan Ross | For Darwin, I'd say it's hard to say | 17:30:50 |
emily | which is sort of what the tier 1 description implies | 17:31:02 |
emily | and will probably never happen, but if we're talking hyphtheticals | 17:31:21 |
Alyssa Ross | In reply to @emilazy:matrix.org I don't think that would be the case if half of all users and contributors were on Darwin though What specifically? | 17:31:41 |
emily | In reply to @qyliss:fairydust.space What specifically? as in, if we had that critical mass then "block indefinitely on Darwin issues" wouldn't imply expecting Linux users to diagnose/solve things, it'd just imply the Darwin half of the project needing to step in. which would be untenable at present precisely because it's not half | 17:33:32 |
Alyssa Ross | I think that misses that Darwin users can work on Linux stuff too, but not necessarily the other way around. | 17:34:15 |
emily | I think plenty of critical Linux components can break from underneath us in the same way too (but admittedly we have much more of a voice there) | 17:34:20 |
Alyssa Ross | I don't think they can — only the kernel really. | 17:34:37 |
emily | In reply to @qyliss:fairydust.space I think that misses that Darwin users can work on Linux stuff too, but not necessarily the other way around. well, you can imagine a Darwin user with an ideological opposition to running a GPL kernel or something. | 17:34:52 |
Alyssa Ross | Because apart from the kernel we're pure on Linux. | 17:34:59 |
Alyssa Ross | In reply to @emilazy:matrix.org well, you can imagine a Darwin user with an ideological opposition to running a GPL kernel or something. Sure, but do we want to treat that the same way? | 17:35:17 |
emily | (also graphics drivers and a host of runtime services. but I more meant that insofar as you can hold back a ruinous core system update on NixOS you can also just not support the new macOS yet. but I don't claim the two situations are wholly analogous) | 17:36:24 |
emily | In reply to @qyliss:fairydust.space Sure, but do we want to treat that the same way? I don't know, but I think there are gradations. I don't think tier 1 Darwin is going to happen any time soon, but e.g. consider that full participation in the infra team is incompatible with unwillingness to use Darwin even at the current tier. | 17:37:58 |
emily | hence use on NixOS infra really being the biggest dividing line in my eyes | 17:38:17 |