14 Sep 2025 |
Lun | wait msgpack-cxx is providing a msgpack-cxx-config.cmake and not a msgpack-config.cmake 🤔
are they putting two different things in the same repo on different branches
| 22:02:14 |
Lun |
C Library See c_master (branch) C++ Library See cpp_master (branch)
| 22:02:32 |
Lun | probably shouldn't be aliased together but the C one should be bumped to the "6.1.0 C" rev | 22:05:54 |
Lun | or was msgpack always the cpp version too but they renamed the target | 22:09:33 |
emily | msgpack was always the C++ version and they renamed the target, yes. | 22:09:47 |
emily | we have msgpack-c | 22:09:49 |
emily | also yes it's evil that they do that with the repo | 22:10:03 |
emily | guy who overheard 30 seconds of conversation about monorepos and got entirely the wrong idea | 22:10:12 |
emily | cf. https://github.com/msgpack/msgpack-c/commit/ec8895155c66943b4b1586d1fafaf81b61745a79 | 22:10:18 |
emily | before that it was, uh, "msgpackc-cxx"? | 22:10:28 |
Lun | whyy | 22:10:35 |
Lun | it was just msgpack in 3.x | 22:10:50 |
emily | probably because it being "msgpack" was confusing as hell | 22:11:04 |
emily | as we have found out | 22:11:05 |
emily | oh wait oh no | 22:11:51 |
emily | https://github.com/msgpack/msgpack-c/issues/872 | 22:11:56 |
emily | I guess msgpack was both? | 22:12:01 |
emily | yeah so 4.0 split them up. so some things might want the C version. fml. | 22:12:50 |
emily | I'll have to replace the alias with a throw I guess. and figure out what wants what. yuck. | 22:14:34 |
Lun | do you mind going with just substituteInPlaceing the target names in some rocm stuff? i think i can get that working reasonably quickly
looks like they have proper support upstream for the new target names but the commits don't apply on the versions we ship due to a bunch of build churn | 22:15:18 |
Lun | * do you mind going with just substituteInPlaceing the target names in some rocm stuff? i think i can get that working reasonably quickly
looks like they have proper support upstream for the new target names but the commits don't apply on the versions we ship due to a bunch of build system churn | 22:15:27 |
emily | works for me | 22:16:49 |
emily | do you mind PRing that? I don't feel qualified to verify it at all | 22:17:11 |
Lun | oh the 6.4 bump might already be past some of the upstream fixes, i'll apply them together and see if anything still needs patched | 22:21:54 |
emily | is the 6.4 bump ready-ish? | 22:22:46 |
emily | (the main reason I'm doing this is to unblock CMake 4) | 22:22:52 |
Lun | I feel like it's ready on my end | 22:23:22 |
Lun | > Policy CMP0167 is not set: The FindBoost module is removed. Run "cmake
> --help-policy CMP0167" for policy details. Use the cmake_policy command to
> set the policy and suppress this warning.
>
> Call Stack (most recent call first):
> /nix/store/qk0hfyfvqqp59gd05wsaf8nd40cmbav5-msgpack-cxx-7.0.0/lib/cmake/msgpack-cxx/msgpack-cxx-config.cmake:30 (find_dependency)
Looks like msgpack-cxx wants to FindBoost
| 22:23:50 |
emily | I guess it needs to propagate it | 22:24:27 |
emily | pushed | 22:27:32 |