| 14 May 2023 |
John Ericson | Alyssa Ross: oh yeah, duh | 18:05:15 |
John Ericson | but hopefully it is just the first few stages that are handcoded syscalls | 18:05:34 |
Alyssa Ross | yeah, and the hard work is done, just needs translated | 18:06:15 |
John Ericson | and the work is so incremental (port this and that thing) it is very friendly | 18:07:00 |
John Ericson | quick save ftw | 18:07:24 |
emilytrau | John Ericson: i was thinking of a stage0-darwin, but probably is funnier than it is practical 🤣 | 23:31:00 |
John Ericson | yeah darwin is...dfficult | 23:31:35 |
John Ericson | bsd is not so difficult | 23:31:38 |
John Ericson | at least no cat and mouse with Apple changing stuff on us | 23:32:00 |
John Ericson | emilytrau: I am waiting for the build to complete (mes...) but I think I got the parallelism PR working now | 23:33:34 |
John Ericson | It also separates compilers themselves and libs, which is kind of my hobbyhorse :) | 23:33:53 |
John Ericson | err not mes, mes is fast, tcc built with mes | 23:34:33 |
John Ericson | that is slow | 23:34:35 |
John Ericson | I do want to get cross compilation going at some point (hex0 on one platform, hex1 on another, sure, no problem) and the compiler libs split will help with that | 23:34:53 |
John Ericson | CA derivations will allow for a sort of dynamic programming like thing where whenever two paths converge, on memoizes | 23:35:54 |
John Ericson | so instead of num_plats^bootstrapp_steps combinations, there are just num_plats * bootstrapp_steps | 23:36:18 |
John Ericson | (assuming every step can cross compiler) | 23:36:27 |
John Ericson | * (assuming every step can cross compile) | 23:36:29 |
John Ericson | OK build finished, https://github.com/NixOS/nixpkgs/pull/231894 undrafted | 23:42:44 |
John Ericson | I recommmend reviewing with ignoring space changes 😓 | 23:42:56 |
John Ericson | and probably commit-by-commit too | 23:43:04 |
| 15 May 2023 |
John Ericson | Alyssa Ross: do you want to add freebsd to release-cross.nix after all your hard work? | 00:27:43 |
| 16 May 2023 |
misuzu | In reply to @raitobezarius:matrix.org PCIe is something quite minimal we want and is still under review with current patchset kernel doesn't see nvme ssd when it's initialized in u-boot, but if boot partition is on sdcard it works fine. I hope this will be fixed soon | 08:39:15 |
j-k | emilytrau John Ericson anything else in progress for minimal-bootstrap? Am I right in thinking we're on sysa/run-after-fiwix.kaem at the moment? | 18:46:25 |
John Ericson | j-k: I am not working on anything, I did my parallelism bit and am done for the moment | 18:47:12 |
j-k | yeah I pulled the latest copy & it builds very nicely. I'm playing with adding strictDeps to the derivationWithMeta but I imagine how we add extra bits like that will need some discussion
Since it looks like we're on that file I'm probably going to try add tar & gzip in the same vein as gnumake/gnupatch | 18:49:32 |
John Ericson | j-k: strictDeps is a stdenv thing, derivationWithMeta already lacks unstrictDeps! | 19:19:27 |
John Ericson | unless I am missing something | 19:19:35 |
j-k | ah ok got it, I thought it might be a derivation builtin thing. my bad | 19:20:00 |
j-k | For gzip and tar I'm going to need some replacement files & makefiles from live-bootstrap https://github.com/fosslinux/live-bootstrap/tree/1bc4296091c51f53a5598050c8956d16e945b0f5/sysa/gzip-1.2.4/files https://github.com/fosslinux/live-bootstrap/tree/1bc4296091c51f53a5598050c8956d16e945b0f5/sysa/tar-1.12/files
any thoughts on vendoring them in nixpkgs, vs creating .patch files that are equivalent, vs builtins.toFile them in?
| 19:31:56 |