| 1 Jun 2021 |
 LnL | heh | 17:21:57 |
| * LnL continues hacking on his nerves project :D | 17:22:09 |
|  prusnak changed their display name from stick to prusnak. | 18:04:49 |
|  telent joined the room. | 19:22:21 |
 toonn | LnL: So, the bootstrap-tools seem to still build fine. Added a commit which uses lib.getLib for most of the /lib outputs and only tested that on the second bootstrap-tools, i.e., the one with the first passed to stdenv, produced the same output hash so I think it's fine. | 19:52:40 |
| toonn | Next step is to push it to the PR and change the target branch to master? | 19:52:58 |
| LnL | context? | 20:22:45 |
| toonn | The LLVM 11 bump. | 20:23:15 |
| toonn | #121055 | 20:23:21 |
| LnL | oh right, sounds good :D | 20:23:29 |
| LnL | ris_: based on going through the llvm pr I see we already have patches for cmake to remove frameworks | 20:25:58 |
 ris_ | cool | 20:26:27 |
| LnL | if you wanted a precedent for the curl situation, that said using normal options instead having to patch stuff for these cases would be nice | 20:27:57 |
| ris_ | yeah i think for the medium term that sounds like a sensible plan | 20:28:56 |
| ris_ | if curl go on to use frameworks more thoroughly, then that might be time to rethink | 20:29:25 |
| LnL | if I go back I see I made it conditional for bootstrapping initially now it's always just patched out | 20:29:33 |
| LnL | yeah, if my idea for bootstrapping works then it doesn't matter anymore | 20:30:59 |
| LnL | and as an added benefit we kill a bunch of dependencies in the bootstrap tree (except for bootstrap-tools itself) | 20:31:24 |
| LnL | toonn: I see a bunch more stuff in there, those Libsystem changes etc. are needed for the bump? | 20:32:57 |
| toonn | LnL: Yeah, everything but the xnu patch. But that's a bugfix unless my python understanding's lacking. | 20:35:05 |
| toonn | Well, "bug." It doesn't seem to have caused any trouble so far. | 20:35:25 |
| ris_ | one day if i ever understand the various bootstrap loops i'm going to draw a big tube map style diagram to document them | 20:35:44 |
| LnL | alright, just checking I was expecting more package fixes and less stdenv stuff | 20:36:17 |
| LnL | but it's not all that fancy, guess the target things are handled by newer compilers now | 20:36:55 |
| toonn | Here's my notes on the TargetConditionals.h stuff, https://git.io/JGz3q | 20:40:11 |
| toonn | I haven't fixed any packages yet because I haven't built any. | 20:40:25 |
| toonn | I'm not sure how to go about that actually. I thought a hydra jobset was the first step? | 20:41:00 |
| LnL | yeah, unless you already encountered obvious stuff | 20:41:25 |
| LnL | eval is pending... | 20:41:47 |
| toonn | Most of the commits deal with new flags introduced between LLVM 9 and 11, turning them back to the old behavior. | 20:43:01 |
