| 31 Jul 2022 |
@elvishjerricco:matrix.org | I should time the actual make-initrd-ng runtime, because I'm sure a ton of that time is cpio and the compressor | 14:53:43 |
@elvishjerricco:matrix.org | Oh, uh, lul, that's way more dramatic. Old:
real 1.09
user 0.56
sys 0.58
Goblin:
real 0.27
user 0.06
sys 0.20
Notice how much of that time is actual rust code running. Nice
| 15:00:03 |
K900 | Try it without stripping | 15:02:48 |
@elvishjerricco:matrix.org | real 0.09
user 0.00
sys 0.08
lul
| 15:04:44 |
@elvishjerricco:matrix.org | Compared to old with stripping commented out:
real 0.89
user 0.49
sys 0.44
| 15:06:49 |
@linus:schreibt.jetzt | exec is expensive! Especially in the Nix world. | 15:18:44 |
@elvishjerricco:matrix.org | I was originally very happy about rewriting this logic in rust because my bash code took several minutes and this took about a second. Now it's a tenth of a second. That makes me smile | 15:30:28 |
K900 | I just switched my laptop to a goblin-built initrd | 15:58:23 |
K900 | And nothing exploded | 15:58:26 |
K900 | So that's nice | 15:58:29 |
K900 |  Download image.png | 15:59:46 |
K900 | Teehee | 15:59:46 |
K900 | Also, I just realized | 16:03:14 |
K900 | Bootspec dragged in a kernel to my WSL system :( | 16:03:34 |
@elvishjerricco:matrix.org | Is there a way to get the rust dependencies to build in a separate derivation from make-initrd-ng? It's really annoying that what used to be like a 1s build now takes closer to 10s when trying to iterate on it | 18:23:51 |
| 1 Aug 2022 |
| lukegb (he/him) joined the room. | 16:25:23 |
| Winter (she/her) joined the room. | 16:49:25 |
Winter (she/her) | In reply to @elvishjerricco:matrix.org Is there a way to get the rust dependencies to build in a separate derivation from make-initrd-ng? It's really annoying that what used to be like a 1s build now takes closer to 10s when trying to iterate on it Kind of, but our infrastructure doesn't support it. | 16:50:03 |
Winter (she/her) | In reply to @elvishjerricco:matrix.org Is there a way to get the rust dependencies to build in a separate derivation from make-initrd-ng? It's really annoying that what used to be like a 1s build now takes closer to 10s when trying to iterate on it * Yes, but our infrastructure doesn't support it. | 16:50:14 |
Winter (she/her) | (External projects do, though.) | 16:50:54 |
@elvishjerricco:matrix.org | It'd be really nice. haskellPackages has spoiled me :P | 16:52:36 |
Winter (she/her) | K900: Why would the Goblin changes potentially fix cross? Like, were we not able to pass in the proper patchelf binary paths? | 16:53:04 |
@elvishjerricco:matrix.org | Winter (she/her): It looks like patchelf would have done the trick | 16:53:29 |
@elvishjerricco:matrix.org | But there's a lot to like about using goblin for other reasons | 16:53:42 |
Winter (she/her) | Yeah, definitely. | 16:53:49 |
K900 | patchelf does seem to work, but ldd doesn't, and neither does strip | 16:53:53 |
Winter (she/her) | But was just curious about that aspect. | 16:53:53 |
K900 | Linus fixed it to use the right strip | 16:54:06 |
K900 | And I honestly wasn't 100% sure patchelf would have worked | 16:54:17 |
K900 | And also I knew goblin was great | 16:54:23 |