| 4 Jul 2026 |
 K900 | That should work | 16:09:18 |
 hexa | yeah, XM3 here, too | 16:09:29 |
| K900 | I'll get back to you in uhhh several webkits | 16:09:46 |
 emily | fwiw I dropped the -fno-merge-constants because it seems like their AOSP build isn't setting it | 16:10:18 |
| emily | so seemed pointless (and e.g. was already doing nothing on Clang/Darwin) | 16:10:42 |
| K900 | That may have been inherited from a different time | 16:11:13 |
| K900 | Or from AOSP static builds | 16:11:21 |
| K900 | 
Download Well I expected worse | 16:15:28 |
 tom | nibabel fix: https://github.com/NixOS/nixpkgs/pull/538403 | 16:16:51 |
| emily | ffs, Swift on Darwin just ignores -use-ld=lld | 16:20:25 |
 dramforever | fortify requires optimization | 16:26:39 |
| dramforever | wait i can't find it documented anywhere hmm | 16:27:45 |
| emily | that's definitely a thing I know as lore | 16:28:10 |
| emily | but not sure where it's documented | 16:28:14 |
| dramforever | oh it's a #warning in glibc headers | 16:28:37 |
| dramforever | you'll see it if you try to turn on fortify without -O | 16:28:52 |
| K900 | Yeah it yells at you | 16:29:03 |
| K900 | I guess the right question is "why is -O2 only part of hardening and not enabled in other places" | 16:32:37 |
| emily | because build systems handle that and injecting flags behind build systems is evil and we should do less of it rather than more? | 16:33:16 |
| emily | like cc foo.c on the command line should probably not optimize by default… | 16:33:29 |
| emily | > error: install_name_tool: changing install names or rpaths can't be redone for: /nix/store/sy8djyim6q0dc5cadm9f7xfkjqzawp9p-swiftpm-bootstrap-5.10.1/lib/swift/pm/ManifestAPI/libCompilerPluginSupport.dylib (for architecture arm64) because larger updated load commands do not fit (the program must be relinked, and you may need to use -headerpad or -headerpad_max_install_names)
lord…
| 16:36:06 |
| K900 | Wha | 16:36:46 |
| emily | more Swift nonsense | 16:39:29 |
| emily | hexa: do you know what's up with pendulum? | 16:51:28 |
| emily | it's broken on Linux too and it will block e.g. mpv | 16:51:39 |
| emily | it looks like 3.14 to me | 16:52:07 |
| hexa | looks fine on linux to me | 16:52:27 |
| hexa | yean, can subsitute all of aarch64/x86_64 py313/314 | 16:53:17 |
| hexa | and I can see it is also broken on py313 on aarch64-darwin | 16:55:06 |
| emily | hm | 16:55:07 |
