Open mattwthompson opened 5 months ago
@conda-forge-admin, please rerender
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 (recipe
) and found it was in an excellent condition.
I do have some suggestions for making it better though...
For recipe:
{{ stdlib("c") }}
as well. For further details, please see https://github.com/conda-forge/conda-forge.github.io/issues/2102.Reading through https://conda-forge.org/news/2024/03/24/stdlib-migration/
if a feedstock uses a - {{ compiler(...) }} jinja in the build section
that's us
if a feedstock uses - sysroot_linux-64 2.17 # [linux64] (or a variation)
don't think so
if a feedstock sets MACOSX_DEPLOYMENT_TARGET in conda_build_config.yaml
no
In meta.yaml, you can then remove any variations of
no
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 (recipe
) and found it was in an excellent condition.
Is the recipe's use of "a compiler" a hold-over from when NAB was a part of AmberTools? If so, that might be removed now.
I have no idea, but three compilers are listed so I figure it's not an accident. (I also don't know when NAB was removed.)
Just need a rerender.
@conda-forge-admin, please rerender
Hi! This is the friendly automated conda-forge-linting service.
I was trying to look for recipes to lint for you, but it appears we have a merge conflict. Please try to merge or rebase with the base branch to resolve this conflict.
Please ping the 'conda-forge/core' team (using the @ notation in a comment) if you believe this is a bug.
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 (recipe
) and found it was in an excellent condition.
@conda-forge-admin, please rerender
@conda-forge-admin, please rerender
2024-05-08T19:28:11.6566160Z error: /Users/runner/miniforge3/conda-bld/ambertools_1715193701358/_build_env/bin/x86_64-apple-darwin13.4.0-install_name_tool: changing install names or rpaths can't be redone for: /Users/runner/miniforge3/conda-bld/ambertools_1715193701358/_h_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehol/lib/python3.12/site-packages/pytraj/analysis/c_analysis/c_analysis.cpython-312-darwin.so (for architecture x86_64) because larger updated load commands do not fit (the program must be relinked, and you may need to use -headerpad or -headerpad_max_install_names)
2024-05-08T19:28:11.6575240Z
2024-05-08T19:28:12.1381020Z Traceback (most recent call last):
2024-05-08T19:28:12.2383390Z File "/Users/runner/miniforge3/bin/conda-build", line 11, in <module>
2024-05-08T19:28:12.2853650Z sys.exit(execute())
2024-05-08T19:28:12.2964820Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/cli/main_build.py", line 581, in execute
2024-05-08T19:28:12.2992700Z api.build(
2024-05-08T19:28:12.3087000Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/api.py", line 250, in build
2024-05-08T19:28:12.3163640Z return build_tree(
2024-05-08T19:28:12.3211430Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/build.py", line 3762, in build_tree
2024-05-08T19:28:12.3251290Z packages_from_this = build(
2024-05-08T19:28:12.3312040Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/build.py", line 2839, in build
2024-05-08T19:28:12.3347290Z newly_built_packages = bundlers[pkg_type](output_d, m, env, stats)
2024-05-08T19:28:12.3603040Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/build.py", line 1974, in bundle_conda
2024-05-08T19:28:12.3657740Z files = post_process_files(metadata, initial_files)
2024-05-08T19:28:12.3698910Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/build.py", line 1782, in post_process_files
2024-05-08T19:28:12.3764330Z post_build(m, new_files, build_python=python)
2024-05-08T19:28:12.3797000Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/post.py", line 1728, in post_build
2024-05-08T19:28:12.3826850Z post_process_shared_lib(m, f, prefix_files, host_prefix)
2024-05-08T19:28:12.3879100Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/post.py", line 1604, in post_process_shared_lib
2024-05-08T19:28:12.3915810Z mk_relative_osx(path, host_prefix, m, files=files, rpaths=rpaths)
2024-05-08T19:28:12.3953740Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/post.py", line 494, in mk_relative_osx
2024-05-08T19:28:12.3984230Z macho.add_rpath(path, rpath_new, build_prefix=prefix, verbose=True)
2024-05-08T19:28:12.4014090Z File "/Users/runner/miniforge3/lib/python3.10/site-packages/conda_build/os_utils/macho.py", line 299, in add_rpath
2024-05-08T19:28:12.4070290Z raise RuntimeError("install_name_tool failed with exit status %d" % code)
2024-05-08T19:28:12.4532360Z RuntimeError: install_name_tool failed with exit status 1
2024-05-08T19:28:18.9521250Z
2024-05-08T19:28:18.9571970Z ##[error]Bash exited with code '1'.
2024-05-08T19:28:18.9748150Z ##[section]Finishing: Run OSX build
@dacase @njzjz - I'm syncing up with @mattwthompson on his ongoing work, and he's out of ideas on this one. Do either of you have thoughts on how to proceed?
Can someone interpret the failures here? In the past, we have never had a condaforge ambertools package that included MPI. Does it work (as a test) to build a serial package with the AmberTools24 sources? I'm asking because most of the unsuccessful checks above have "mpi" in the failure name (and have "osx" to boot.)
Is there a feasible pathway to build a serial condaforge package now, and continue to work towards including MPI later? Or, only support MPI on Linux? [I've personally more or less given up on OSX for my own work, but I do try to build things on OSX from time to time; getting MPI + OSX to work is often a sticking point.)
Apologies if I am way off base here about what is going wrong. But if I had a better idea of what was really going wrong, I might be able to ask Tim Giese (who is much cleverer than I am) to take look.
MPI support was added in this PR: https://github.com/conda-forge/ambertools-feedstock/pull/133
It looks like the problem is macOS + NumPy 1.26 (notice that osx_64_mpinompinumpy1.26python3.12
is failing, but all other combinations with MPI seem to be building).
I'm not sure what would cause this failure nor have I looked into it: https://numpy.org/doc/stable/release/1.26.0-notes.html#numpy-1-26-0-release-notes
The error message shows there are some problems with pytraj/analysis/c_analysis/c_analysis.cpython-312-darwin.so
.
I see some other feedstocks have similar issues: https://github.com/search?q=org%3Aconda-forge+changing+install+names+or+rpaths&type=code
This might work: https://github.com/conda-forge/siesta-feedstock/blob/6092b241ed7f0d6921a5f2ae10e94cc1183bd0b2/recipe/build.sh#L53-L58
A bit of a rewording of a previous question:
Is it possible to build a limited release of AmberTools24, with a hope of improving it in future dot releases?
For example, pytraj is available as a separate github package, and (as I understand it) is not really needed by most people who would install the conda version of AmberTools24. (Also, Hai Nyguen has been putting lots of changes into the github version of pytraj which are not in the AmberTools release; so we might consider leaving pytraj out of AmberTools24 altogether.)
I'm not sure what else is holding things up, but we could consider those as well.
This is of course not an ideal solution. But it seems worth trying balance competing interests here: conda users now are missing out on a lot of improvements in AmberTools24, versus
some downside of having dot versions of AmberTools24 that differ in functionality.
Making a partial release is certainly an option, as is splitting out the programs within AmberTools into individual packages so people who need only one tool don't need to deal with issues of other tools (this was brought up in https://github.com/conda-forge/ambertools-feedstock/issues/119)
@conda-forge-admin, please rerender
Hi! This is the friendly automated conda-forge-linting service.
I was trying to look for recipes to lint for you, but it appears we have a merge conflict. Please try to merge or rebase with the base branch to resolve this conflict.
Please ping the 'conda-forge/core' team (using the @ notation in a comment) if you believe this is a bug.
@conda-forge-admin, please rerender
I might need to re-base this PR by dropping all the renderings and re-running
Checklist
0
(if the version changed)conda-smithy
(Use the phrase code>@<space/conda-forge-admin, please rerender in a comment in this PR for automated rerendering)