-
Notifications
You must be signed in to change notification settings - Fork 44
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
Update spack fork from spack develop 2022/12/22 #424
Update spack fork from spack develop 2022/12/22 #424
Conversation
@AlexanderRichert-NOAA Do you understand this error?
The code doesn't have two |
No. I pushed a fix based on the function name typo, but I'm not sure what's going on with the double "@" in the output there. I'm hoping it's just a bug in the error output... |
If I had read the error more carefully ... There are new CI failures for Linux, apparently because apparenty something requires a Python newer than 3.9 and the external we have on Linux CI (3.8). I'll check and maybe switch to spack-built python, wanted to do that a while ago already. |
The "satisifes" typo is the one I was referring to-- the fix has already been merged. Let me know if you want me to look at the Python/CI matter. Which images are using 3.8? |
I'm already testing the Linux build with internal Python. It's a ubuntu20 image I think. The update of the CI system to use AWS codebuild with more powerful, faster and better controllable images is long planned, just waiting for the spack repository move to happen. |
But, fwiw, I pulled the spack develop branch this morning, and it didn't have the bacio fix? |
https://github.com/spack/spack/tree/develop/var/spack/repos/builtin/packages/bacio/package.py |
Guess I was 1hr too early !
… On Dec 22, 2022, at 2:26 PM, Alex Richert ***@***.***> wrote:
But, fwiw, I pulled the spack develop branch this morning, and it didn't have the bacio fix?
https://github.com/spack/spack/tree/develop/var/spack/repos/builtin/packages/bacio/package.py <https://github.com/spack/spack/tree/develop/var/spack/repos/builtin/packages/bacio/package.py>
spack/spack#34663 <spack/spack#34663>
—
Reply to this email directly, view it on GitHub <#424 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB5C2RJJ5HEH7FODSE7XEF3WOTBPTANCNFSM6AAAAAATG4KMYI>.
You are receiving this because you authored the thread.
|
@AlexanderRichert-NOAA - I guess the new logic in bacio isn't working as expected. I'll go back to what we had in our spack fork until there is a proper fix in the authoritative repo?
|
Oof. Looks like a Python bug. Depending on the Python version, the code
gives different results. 3.6.8 gives "2.4.1", 3.8.12 and 3.10.5 give "2.4.12.4.1". Must be matching a beginning or end of line. I think the fix will be to change "*" to "+". |
Let me try
… On Dec 22, 2022, at 3:23 PM, Alex Richert ***@***.***> wrote:
@AlexanderRichert-NOAA <https://github.com/AlexanderRichert-NOAA> - I guess the new logic in bacio isn't working as expected. I'll go back to what we had in our spack fork until there is a proper fix in the authoritative repo?
'/usr/local/bin/cmake' '-G' 'Unix Makefiles' '-DCMAKE_INSTALL_PREFIX:STRING=/home/runner/work/spack-stack/spack-stack/envs/default/install/intel/2021.8.0/bacio-2.4.1-pjld7qk' '-DCMAKE_BUILD_TYPE:STRING=RelWithDebInfo' '-DBUILD_TESTING:BOOL=OFF' '-DCMAKE_INTERPROCEDURAL_OPTIMIZATION:BOOL=OFF' '-DCMAKE_VERBOSE_MAKEFILE:BOOL=ON' '-DCMAKE_INSTALL_RPATH_USE_LINK_PATH:BOOL=ON' '-DCMAKE_INSTALL_RPATH:STRING=/home/runner/work/spack-stack/spack-stack/envs/default/install/intel/2021.8.0/bacio-2.4.1-pjld7qk/lib;/home/runner/work/spack-stack/spack-stack/envs/default/install/intel/2021.8.0/bacio-2.4.1-pjld7qk/lib64' '-DCMAKE_PREFIX_PATH:STRING=' '-DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON' '/home/runner/work/spack-stack/spack-stack/cache/build_stage/spack-stage-bacio-2.4.1-pjld7qkpid7oxovqaxhbi7auxsip6gd5/spack-src'
1 error found in build log:
>> 3 CMake Error at CMakeLists.txt:5 (project):
4 VERSION "2.4.12.4.1" format invalid.
5
6
7 -- Configuring incomplete, errors occurred!
Oof. Looks like a Python bug. Depending on the Python version, the code
import re
print(re.sub(".*","2.4.1","2.4.1"))
gives different results. 3.6.8 gives "2.4.1", 3.8.12 and 3.10.5 give "2.4.12.4.1". Must be matching a beginning or end of line. I think the fix will be to change "*" to "+".
—
Reply to this email directly, view it on GitHub <#424 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB5C2RPRXQNJ7ZXXYPPRWL3WOTIFPANCNFSM6AAAAAATG4KMYI>.
You are receiving this because you authored the thread.
|
If that works, you can cherry-pick 36caf0a390557a49073097b40486b243e422b7e2 and update bacio in the authoritative repo. This way we won't have merge conflicts down the road. |
…o feature/update_spack_from_authoritative_20221222
517ee99
to
3421fe7
Compare
.gitmodules
Outdated
##branch = develop | ||
#url = https://github.com/NOAA-EMC/spack | ||
#branch = jcsda_emc_spack_stack | ||
url = https://github.com/climbfuji/spack |
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.
Is this intentional? Seems like you are switching to your fork here?
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.
That's needed for the CI tests to work properly. Until the spack PR is merged, the spack submodule pointer in spack-stack must point to the branch and fork to find the updated code. Procedure is: CI tests need to complete, then we merge the spack PR (pending approval, of course), then update the spack
submodule pointer in spack-stack and revert .gitmodules
. Then CI runs again, but can be bypassed if the code in the spack PR is identical with the final code in our (EMC) spack fork/branch - this is normally the case and can be assessed quickly (I do that after merging the spack PR).
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.
BTW, the ufs-weather-model is using the same for its code commit procedures.
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.
cool
…o feature/update_spack_from_authoritative_20221222
#WIP
Description
Updates to documentation, CI builds and package configurations (in
config/common/packages.yaml
and in container configs) to work with the updated spack code from JCSDA/spack#210. Important updates are:py-scipy
from1.8.0
to1.9.3
to fix build errors and duplicate packagesNote: For now, keep the Homebrew Python for macOS CI runs. We should change this (and the Linux instructions for local systems) to spack-built python in the near future. And consider doing the same on HPCs.
Update .gitmodules and submodule pointer for spack for code review and testing
Issues
Working towards #394 and #376.
Dependencies
Testing
skylab-dev
template +jedi-tools-env
)