| 22 May 2025 |
 emily | we have mechanisms that basically nobody understands and that cause all kinds of bad and confusing behaviour, and in return we get a cross-compilation interface that most people don't understand and can't use correctly and have a harder time with than if we just had explicit package sets and monolithic dep lists. yay? | 08:13:27 |
 Alyssa Ross | Yeah... | 08:15:50 |
 Tristan Ross | After https://github.com/NixOS/nixpkgs/pull/365057, I plan on working on https://github.com/NixOS/nixpkgs/pull/376197 and from there, I hope to redo some of lib.systems.elaborate. I've shared with Alyssa my plan, it's to make it use the option system so when we work on what's described in https://github.com/NixOS/nixpkgs/pull/400351#issuecomment-2888530195, we wouldn't have to duplicate things. Likely would have lib/systems/options.nix to describe that and lib.systems.elaborate in lib/systems/default.nix would just handle actually using the options and applying them. | 16:41:46 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/pull/409851 is making it start to become possible to replace things more. I think we need something like this for cxxlib, rtlib, unwinderlib, etc. | 19:46:16 |
| 23 May 2025 |
 aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | In reply to @rosscomputerguy:matrix.org
After https://github.com/NixOS/nixpkgs/pull/365057, I plan on working on https://github.com/NixOS/nixpkgs/pull/376197 and from there, I hope to redo some of lib.systems.elaborate. I've shared with Alyssa my plan, it's to make it use the option system so when we work on what's described in https://github.com/NixOS/nixpkgs/pull/400351#issuecomment-2888530195, we wouldn't have to duplicate things. Likely would have lib/systems/options.nix to describe that and lib.systems.elaborate in lib/systems/default.nix would just handle actually using the options and applying them. Would transforming lib.systems.elaborate adds eval overhead? I found even when I add gccarch condition to canExecute, it increased the eval time by 0.2%, if not noise | 03:08:03 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | * Would transforming lib.systems.elaborate add eval overhead? I found even when I added gccarch condition to canExecute, it increased the eval time by 0.2%, if not noise | 03:08:24 |
| Tristan Ross | I'm not sure, we'd have to benchmark it | 03:08:41 |
| 24 May 2025 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/pull/410056 is ready. ARM trusted firmware works and with an easy patch, Firefox works. | 02:30:09 |
| Tristan Ross | Redacted or Malformed Event | 02:30:32 |
| Tristan Ross | Redacted or Malformed Event | 02:30:50 |
| Tristan Ross | Redacted or Malformed Event | 02:31:22 |
| Tristan Ross | Redacted or Malformed Event | 02:31:44 |
|  SigmaSquadron joined the room. | 02:38:34 |
|  uep joined the room. | 03:34:59 |
| Tristan Ross | Redacted or Malformed Event | 03:38:20 |
| uep | wat | 03:38:45 |
| Tristan Ross | Redacted or Malformed Event | 03:38:52 |
 SomeoneSerge (hash-versioned python modules when) | In reply to @emilazy:matrix.org
since foo.dev is to be consumed on buildPlatform We select .dev for buildInputs too, and use it to propagate lib and static which are for hostPlatform | 09:27:25 |
| emily | right. (I think we're agreeing?) | 09:27:54 |
| Tristan Ross | In reply to @rosscomputerguy:matrix.org
https://github.com/NixOS/nixpkgs/pull/410056 is ready. ARM trusted firmware works and with an easy patch, Firefox works. Ok now that mjolnir isn't removing my every single message, I can link the patch which moves Firefox to using cross stdenvs. https://gist.github.com/RossComputerGuy/525eb9f31ecee52c40338bd0af8cb513 | 13:18:21 |
 Randy Eckenrode | I’d be curious whether that works with building Wine (particularly on Darwin without the hook). | 14:01:30 |
| emily | how much cheaper is instantiating a cross stdenv to instantiating an entire Nixpkgs? | 14:04:50 |
| Randy Eckenrode | Not sure about DXVK. That’s written as a derivation for Windows. dxvk_2 even supports native builds on Linux and FreeBSD, so I wouldn’t want to break that. | 14:04:53 |
| emily | I thought most of the expense was precisely in evaluating stdenv and all its dependencies | 14:05:00 |
| Randy Eckenrode | NIX_SHOW_STATS=1 can be used to compare eval performance with and without. | 14:06:09 |
| Randy Eckenrode | * NIX_SHOW_STATS=1 can be used to compare eval performance with and without the change. | 14:11:54 |
| Tristan Ross | In reply to @emilazy:matrix.org
how much cheaper is instantiating a cross stdenv to instantiating an entire Nixpkgs? I'm guessing a decent amount since pkgs* comes through stage.nix and to get there, the config has to be evaluated. Since you're creating a new nixpkgs instance, you're evaluating that twice. Once for the root and again for the new instance. It also has to evaluate the native stage again and then inject the cross stages. So I'd say there's a good amount but I haven't gotten down to checking exactly how much yet. | 17:54:16 |
| Tristan Ross | In reply to @reckenrode:matrix.org
I’d be curious whether that works with building Wine (particularly on Darwin without the hook). Theoretically, it should | 17:55:25 |
| Tristan Ross | We have this at least https://github.com/NixOS/nixpkgs/actions/runs/15222517519?pr=410056 | 17:56:12 |
| Tristan Ross | I'm curious what eval performance would look like with moving absolutely everything over. Though, I don't want to move too much in this PR. | 17:57:16 |
