pinning the pymatgen version?
unkcpz opened this issue · comments
Install the package directly will install pymatgen==2023.7.14 since the version is not pinned.
I got:
ValueError: Error occurred validating port 'inputs.metadata.options.parser_name': invalid parser specified: Failed to load entry point 'opsp.pseudo.oncv':
Traceback (most recent call last):
File "/opt/conda/lib/python3.9/site-packages/aiida/plugins/entry_point.py", line 237, in load_entry_point
loaded_entry_point = entry_point.load()
File "/opt/conda/lib/python3.9/site-packages/importlib_metadata/__init__.py", line 207, in load
module = import_module(match.group('module'))
File "/opt/conda/lib/python3.9/importlib/__init__.py", line 127, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "<frozen importlib._bootstrap>", line 1030, in _gcd_import
File "<frozen importlib._bootstrap>", line 1007, in _find_and_load
File "<frozen importlib._bootstrap>", line 986, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 680, in _load_unlocked
File "<frozen importlib._bootstrap_external>", line 850, in exec_module
File "<frozen importlib._bootstrap>", line 228, in _call_with_frames_removed
File "/home/aiida/aiida-opsp/aiida_opsp/parser.py", line 8, in <module>
from abipy.ppcodes.oncv_parser import OncvParser
File "/opt/conda/lib/python3.9/site-packages/abipy/__init__.py", line 18, in <module>
from abipy.core import release
File "/opt/conda/lib/python3.9/site-packages/abipy/core/__init__.py", line 4, in <module>
from .structure import *
File "/opt/conda/lib/python3.9/site-packages/abipy/core/structure.py", line 27, in <module>
from abipy.flowtk import PseudoTable
File "/opt/conda/lib/python3.9/site-packages/abipy/flowtk/__init__.py", line 15, in <module>
from .tasks import *
File "/opt/conda/lib/python3.9/site-packages/abipy/flowtk/tasks.py", line 37, in <module>
from .abitimer import AbinitTimerParser
File "/opt/conda/lib/python3.9/site-packages/abipy/flowtk/abitimer.py", line 2, in <module>
from pymatgen.io.abinit.abitimer import AbinitTimerParserError, AbinitTimerParser, AbinitTimerSection
ImportError: cannot import name 'AbinitTimerParserError' from 'pymatgen.io.abinit.abitimer' (/opt/conda/lib/python3.9/site-packages/pymatgen/io/abinit/abitimer.py)
This is related to the following change:
materialsproject/pymatgen@5b88fc1
In the latest abipy version the import has been updated to fix this:
abipy/abipy/flowtk/abitimer.py
Lines 1 to 3 in d4dd68e
But if your environment still uses an older pymatgen version due to other package dependency restrictions, this will fail again, so I suggest an import based on the pymatgen version.
Pinning the pymatgen version is also not a good idea for this reason. I think if everyone starts pinning their pymatgen versions we're going to run into dependency resolution conflicts.
Thanks @mbercx, I think I have to not pinning pymatgen, but this means if they change the API in the future again, things break.
Yeah, I know... But if we pin it every package that uses it in the common workflows will have to pin it to the same version, else we can't install aiida-common-workflows at all. We could define a range and update the upper limit from time to time?