| 24 Jun 2025 |
|  sielicki joined the room. | 22:17:05 |
| 25 Jun 2025 |
 emily | tomberek: John Ericson: in re relocatable store https://sourceware.org/bugzilla/show_bug.cgi?id=31959 | 15:13:35 |
| emily | does anyone understand better than me why we have both .parsed.cpu.arch and .gcc.arch | 17:45:54 |
| emily | like I get that the latter can be a more specific model but we also condition on its existence in canExecute in a way I don't fully understand, and set it for some systems even when it's identical | 17:58:31 |
 K900 | gcc.arch is -march I think? | 17:59:43 |
| emily | right. but why do we set it to armv8-a explicitly for AArch64? | 19:02:12 |
| emily | that's already how the CPU parses | 19:02:24 |
| K900 | idk probably eval somewhere | 19:02:31 |
| emily | I doubt it. x86 doesn't get the field set | 19:02:54 |
| emily | I suspect it's just a historical thing | 19:03:13 |
 Grimmauld (any/all) | oh btw, did the i686 time thing happen yet? | 19:04:25 |
| K900 | Nope | 19:04:48 |
| K900 | Fuck | 19:04:50 |
| Grimmauld (any/all) | XD | 19:04:56 |
| K900 | I might be sniped into doing it tomorrow :( | 19:04:59 |
| Grimmauld (any/all) | i tried, saw i was totally out of my depth, and gave up after a few hours | 19:05:28 |
| emily | it should really only take like five lines | 19:06:58 |
| K900 | dooo eeeet | 19:07:26 |
| emily | do you want me to do that or test BPI kernels? :p | 19:07:51 |
| Grimmauld (any/all) | you said that last time, and then i got stuck for a day | 19:08:09 |
| emily | you had it in the wrong file I believe | 19:08:26 |
 Tristan Ross | In reply to @emilazy:matrix.org
right. but why do we set it to armv8-a explicitly for AArch64? I believe it's because of the difference between generic and armv8-a | 19:15:19 |
| Tristan Ross | I do plan on reworking the CPU model stuff after the toolchain attributes PR is merged. I don't like that it's called gcc.arch. | 19:17:14 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/pull/365057 almost half a year old now | 19:27:20 |
| emily | I will hopefully have time this weekend to catch up on my review backlog. (there's Darwin stuff I need to get to as well) | 19:32:49 |
| emily | at a glance cxxrtlib is still wrong though | 19:32:54 |
| emily | we don't currently use libcxxrt anywhere I don't think, certainly not on Linux | 19:33:21 |
| emily | (FreeBSD probably should but currently doesn't) | 19:33:27 |
| Tristan Ross | In reply to @emilazy:matrix.org
we don't currently use libcxxrt anywhere I don't think, certainly not on Linux Huh, what should the value be then? | 23:07:27 |
| emily | depends on the platform – but libstdc++ is coupled to libsupc++ for RTTI/exceptions and I think we use libstdc++ on most platforms right now | 23:13:26 |
