Has anyone tried a Python 3.11 conda package build yet?

[swails]

Attempts to do this for AmberTools 23 are failing.

build log here: https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=700933&view=logs&j=9a864fd9-6c8f-52ca-79ce-2aa6dca1a1de&t=10fc5aa2-324e-5982-4c88-6b31fcab16b3)

PR here: conda-forge/ambertools-feedstock#113

[hainm]

I have not tried with python 3.11 yet.

pytraj/trajectory/frame.cpp:45333:1: error: too many initializers for 'PyTypeObject' {aka '_typeobject'}

frame.cpp file is generated by cython and that file is included in the release. wondering if we can have better luck with latest cython version?

[DrDomenicoMarson]

Hi,
I can confirm that I have the same issue building amber23 in an Anaconda environment with python 3.11.3 + cython 0.29.35.

Is there some workaround that can be done to successfully build pytraj? I'd hate to have to disable its installation!

[swails]

Update cython and recythonize the files?

I haven't had a chance to try it out yet.

[DrDomenicoMarson]

Unfortunatelly I don't know how to do that (I'm not compiling from pytraj git repo, but from AmberTools23).

To briefly describe what I've done:

• I created a conda environment with python3.11

• added cython, numpy, mpi4py, and other dependencies

• ran cmake (python is successfully found)

• when running make install the above errors stop the installation, while if I exclude pytraj from the required packages, everything is compiled successfully!

Doesn't the cmake procedure cythonize the files with the available cython?

[hainm]

Please: - force re-cythonizing by updating scripts/base_setup.py to change is_release from True to False And continue to rebuild ambertools Let me know if this works. I am on vacation so I don’t have a chance to try. Hai

[DrDomenicoMarson]

Hi! I'm sorry to bother you on vacation!

I tried as you suggested, but the same error

pytraj/core/c_dict.cpp:460:62: error: invalid use of incomplete type 'PyFrameObject' {aka 'struct _frame'}

unfortunatelly arise...

I checked the make output, and indeed the is_release seems to be recognized. I just paste here the relevant part of the make output, because another warning is also raised:

[ 79%] Building pytraj native library
sys.argv ['setup.py', 'build', '-b', '/opt/amber22_src/build/AmberTools/src/pytraj/CMakeFiles/python-build']
Original compiler envs
CXX = /opt/anaconda3/envs/p11/bin/g++
CC = /opt/anaconda3/envs/p11/bin/gcc
COMPILER = 
use_pip = False, cpptraj_home = 
compile_c_extension = True
2.0.6.dev0
use_prebuilt_cythonized_files =  False
[ 82%] Built target sander_base_obj_omp
pytraj is inside AMBERHOME
using environment: CC=/opt/anaconda3/envs/p11/bin/gcc, CXX=/opt/anaconda3/envs/p11/bin/g++
using CC=/opt/anaconda3/envs/p11/bin/gcc, CXX=/opt/anaconda3/envs/p11/bin/g++
Final compiler envs
CXX = /opt/anaconda3/envs/p11/bin/g++
CC = /opt/anaconda3/envs/p11/bin/gcc
COMPILER = 
install = True
set rpath to /opt/amber22_src/build/AmberTools/src/cpptraj/src
/opt/anaconda3/envs/p11/lib/python3.11/site-packages/setuptools/dist.py:755: SetuptoolsDeprecationWarning: Invalid dash-separated options
!!

        ********************************************************************************
        Usage of dash-separated 'description-file' will not be supported in future
        versions. Please use the underscore name 'description_file' instead.

        By 2023-Sep-26, you need to update your project and remove deprecated calls
        or your builds will no longer be supported.

        See https://setuptools.pypa.io/en/latest/userguide/declarative_config.html for details.
        ********************************************************************************

!!
  opt = self.warn_dash_deprecation(opt, section)
running build
running build_py
copying pytraj/version.py -> /opt/amber22_src/build/AmberTools/src/pytraj/CMakeFiles/python-build/lib.linux-x86_64-cpython-311/pytraj
running build_ext
skipping 'pytraj/analysis/c_analysis/c_analysis.cpp' Cython extension (up-to-date)
skipping 'pytraj/analysis/c_action/actionlist.cpp' Cython extension (up-to-date)
skipping 'pytraj/analysis/c_action/c_action.cpp' Cython extension (up-to-date)
skipping 'pytraj/math/cpp_math.cpp' Cython extension (up-to-date)
skipping 'pytraj/topology/topology.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/topology_objects.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/coordinfo.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/parameter_types.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/c_dict.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/c_options.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/box.cpp' Cython extension (up-to-date)
skipping 'pytraj/core/c_core.cpp' Cython extension (up-to-date)
skipping 'pytraj/utils/cyutils.cpp' Cython extension (up-to-date)
skipping 'pytraj/datafiles/datafiles.cpp' Cython extension (up-to-date)
skipping 'pytraj/datasets/c_datasetlist.cpp' Cython extension (up-to-date)
skipping 'pytraj/datasets/c_datasets.cpp' Cython extension (up-to-date)
skipping 'pytraj/datasets/cast_dataset.cpp' Cython extension (up-to-date)
skipping 'pytraj/trajectory/frame.cpp' Cython extension (up-to-date)
skipping 'pytraj/trajectory/c_traj/c_trajout.cpp' Cython extension (up-to-date)
skipping 'pytraj/trajectory/c_traj/c_trajectory.cpp' Cython extension (up-to-date)
building 'pytraj.core.c_dict' extension

[hainm]

Um, I will give this a try when I gave access to my laptop. Cheers.

[hainm]

On Fri, Jun 9, 2023 at 3:04 PM Hai Nguyen ***@***.***> wrote: Um, I will give this a try when I gave access to my laptop. Cheers.

I don’t have a chance to try amber yet but the standalone pytraj compiled fine with python 3.11 and cython 0.29.x So I suggest to also delete all existing cythonized files .cpp in pytraj and then build again. And sorry for all the troubles. Hai

…

[DrDomenicoMarson]

Just a quick update, deleting all the *.cpp files in the src/pytraj/pytraj directory worked fine with python3.11!

(I also had scripts/base_setup.py with is_release setted to False)

[hainm]

Great. Thanks.

[hainm]

I think we can close this case since the question is answered. #1632 (comment)

Further discussions are happening in https://github.com/conda-forge/ambertools-feedstock