-
-
Notifications
You must be signed in to change notification settings - Fork 30
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Rebuild for abseil_cpp20220623 (redux) #138
Conversation
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
So both the previous(ly working) as well as the new abseil build fail with something java-related:
|
In this PR I still left |
I think the likely cause for the java failure is actually the way we pass the commands to the custom toolchain... |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
regardless of the java thing, this should be merged sooner rather than later. Thanks!!
I say this because somehow the pinning allows for different abseils to exist in the host_env and build_env, see logs in e.g. conda-forge/tensorflow-feedstock#264 and conda-forge/jaxlib-feedstock#130
I don't have a systematic answer to this (and it's definitely just my lack of insight and not necessarily a shortcoming), but I've seen such things before as well. Different deps in the different environments might lead to different resolutions vis-à-vis unfinished migrations. The solution is to specify the respective dep (in this case abseil) in the build env as well, so it gets populated by the same value from the pinning. Introducing this might still run into resolution errors if not all deps of the build env are migrated yet |
I do remember seeing arcane build options like |
I am not sure it is problem per se... In both cases above, things look fine with mixing the host and build. conda-forge/jaxlib-feedstock#130 looks completely fine. I will test the packages more thoroughly later locally before merging. Also the current error in conda-forge/tensorflow-feedstock#264 looks likely to be about missing flatbuffers in run (not really declared as a run dependency upstream, so idk what's going on) |
@conda-forge/bazel any outstanding issues here? Could we merge this soon? I am blocking both conda-forge/jaxlib-feedstock#130 and conda-forge/tensorflow-feedstock#264 waiting for this. They both can pass with the bazel built with older abseil, but I think that's dodgy, and we should build with the newer bazel. I do not have appetite to rebuild them again later due to some bizarre missing symbols (like we always end up doing...) I am blocking them by adding abseil-cpp to build, and so it must satisfy that abseil-cpp and bazel are built with the same pin. Interestingly, tensorflow no longer builds successfully with the older abseil, see detailed errors in conda-forge/tensorflow-feedstock#264 and conda-forge/tensorflow-feedstock#260 (wildly enough, it fails after it is done compiling). Again, could get around this by taking abseil-cpp out of the build section, but I am afraid this is a recipe for disaster... and I'd rather wait for this to go through. |
@conda-forge-admin, please rerender |
We can have a go at unpinning jdk later if you want. It will require substantial edits to to build.sh to restructure the bootstrapping process. |
…nda-forge-pinning 2022.08.23.01.05.47
Hi! This is the friendly conda-forge automerge bot! I considered the following status checks when analyzing this PR:
Thus the PR was passing and merged! Have a great day! |
Bot had a resolution error; tackle main before taking on #134/#135
Closes #141
Closes #145