| 30 Jun 2024 |
|  shekhinah changed their display name from shekhinah to sabaoth. | 15:09:49 |
 lromor | SomeoneSerge (UTC+3): feel free to contact me here in case you need a more "realtime" discussion :) | 18:49:05 |
| lromor | * SomeoneSerge (UTC+3): feel free to contact me here in case you need a more "realtime" discussion. | 18:49:15 |
| lromor | * SomeoneSerge (UTC+3): I answered your github question, feel free to contact me here in case you need more info. | 18:49:40 |
 SomeoneSerge (back on matrix) | In reply to @ss:someonex.net
I think this only really breaks jax and tensorflow, which are broken anyway 🤔 Jax builds, looking into tf ✅ | 21:51:48 |
| 2 Jul 2024 |
|  Jonas Chevalier joined the room. | 10:08:21 |
| Jonas Chevalier | 👋 thought I would join in on the fun for a bit | 10:24:19 |
 hexa (UTC+1) | Great having you here! Feel free to enable the (unfun) config.cudaSupport flag at any time now 🙂 | 10:41:14 |
| hexa (UTC+1) | Wondering what it would take to build and ship unfree redistributable things | 10:41:57 |
| hexa (UTC+1) | The discussion was running in #platform-governance:nixos.org until the Zulip happened 🫠| 10:42:20 |
| lromor | In reply to @zimbatm:numtide.com
👋 thought I would join in on the fun for a bit Someone of numtide developed nix-ld | 10:42:46 |
| lromor | * Someone of numtide developed nix-ld am I right? | 10:42:57 |
| hexa (UTC+1) | that would be Mic92 | 10:43:00 |
| lromor | Nice | 10:43:06 |
| Jonas Chevalier | In reply to @hexa:lossy.network
Wondering what it would take to build and ship unfree redistributable things I'd love to do that | 11:41:34 |
| Jonas Chevalier | In reply to @lromor:matrix.org
Nice Yes, and also https://github.com/numtide/nix-gl-host which might be relevant here | 11:42:19 |
| Jonas Chevalier | For a while I was running CI on all the unfree packages on https://github.com/numtide/nixpkgs-unfree, but it got really expensive, because my CI doesn't cache failed builds, so it would rebuild broken tensorflows and friends over and over again. | 11:55:19 |
| Jonas Chevalier | I have been reading the tensorflow packaging and friends for the past few days, and gained a lot of appreciation for the work that happened there. This goes in the leaderboard of most difficult packaging to work on. | 12:02:51 |
| hexa (UTC+1) | absolutely does. also the reason we are lagging behind on the tensorflow source-built package unfortunately. | 12:17:26 |
| Jonas Chevalier | yeah, it looks like the hard bit isn't the lack of hardware but the amount of combinatorial manpower required. | 12:24:42 |
| Jonas Chevalier | i'd love to throw hardware at that problem :) | 12:24:59 |
| Jonas Chevalier | are the cases where you would use the *WithCuda packages instead of import nixpkgs { config.cudaSupport = true; }? This could create a similar situation as I had when trying to mix torch and tensorflow in one python.withPackages, where you get package name collisions. | 12:50:18 |
| SomeoneSerge (back on matrix) | Holaaaa | 14:55:06 |
| SomeoneSerge (back on matrix) | In reply to @zimbatm:numtide.com
are the cases where you would use the *WithCuda packages instead of import nixpkgs { config.cudaSupport = true; }? This could create a similar situation as I had when trying to mix torch and tensorflow in one python.withPackages, where you get package name collisions. Honestly, not many I think. "A standalone executable that has no transitive dependencies that require enabling optional cuda support". E.g. I think python3Packages.torchWithCuda would link against the wrong ucc/ucx/openmpi unless config.cudaSupport is set | 15:00:46 |
| SomeoneSerge (back on matrix) | I think they were mainly introduced to be put in passthru.tests and for nixpkgs-reviews | 15:01:32 |
| SomeoneSerge (back on matrix) | What I think we need instead is we need to somehow [more] publicly expose nixpkgsFun so we can access it from all-packages.nix and prepare the pkgsCuda and pkgsRocm attributes | 15:03:14 |
| SomeoneSerge (back on matrix) | * What I think we need instead is we need to somehow [more] publicly expose nixpkgsFun so we can access it from all-packages.nix and prepare the pkgsCuda and pkgsRocm attributes (even without recurseIntoAttrs) | 15:03:33 |
| SomeoneSerge (back on matrix) | In reply to @zimbatm:numtide.com
I'd love to do that We'd love that too 🙃 we talk about it but never gain enough momentum to act | 15:05:16 |
| Jonas Chevalier | We could use the Hydra instance in nix-community. It's already there and available. | 16:19:24 |
| Jonas Chevalier | Something like this: https://github.com/nix-community/infra/pull/1335 | 16:45:30 |
