Found by a Codex global repository scan of deepmodeling/dpdispatcher at commit 98a9e08.
Problem
Python support and dependency metadata are inconsistent across PyPI metadata, conda metadata, and CI. pyproject.toml requires dargs>=0.5.0 and requires-python >=3.7; uv run cannot resolve the project because that range includes Python 3.7.0-3.7.3 while dargs depends on typeguard>=4, which requires at least Python 3.7.4. Conda metadata still advertises Python >=3.6 and omits runtime dependencies present in pyproject.toml.
Relevant code
|
dependencies = [ |
|
'paramiko', |
|
'dargs>=0.5.0', |
|
'requests', |
|
'tqdm>=4.9.0', |
|
'typing_extensions; python_version < "3.7"', |
|
'pyyaml', |
|
'tomli >= 1.1.0; python_version < "3.11"', |
|
] |
|
requires-python = ">=3.7" |
|
- python >=3.6 |
|
- pip |
|
- setuptools_scm |
|
- dargs |
|
- paramiko |
|
- requests |
|
- tqdm |
|
|
|
run: |
|
- python >=3.6 |
|
- dargs |
|
- paramiko |
|
- requests |
|
- tqdm |
|
python-version: |
|
- 3.7 |
|
- 3.9 |
|
- '3.10' |
|
- '3.11' |
|
- '3.12' |
Observed locally
uv run --no-dev python -c 'import sys; print(sys.version)' fails with an unsatisfiable dependency split for python_full_version >= '3.7' and < '3.7.4'.
Impact
New development environments can fail to resolve. Conda packages can miss imports such as pyyaml and tomli on older Python versions. CI also skips Python 3.8 while the package classifier declares support.
Suggested fix
Align the supported Python floor and CI matrix with dependency reality, for example requires-python >=3.7.4 or a higher version if intended. Update conda runtime dependencies to include pyyaml and conditional tomli, and align conda's Python requirement with PyPI metadata.
Found by a Codex global repository scan of deepmodeling/dpdispatcher at commit 98a9e08.
Problem
Python support and dependency metadata are inconsistent across PyPI metadata, conda metadata, and CI.
pyproject.tomlrequiresdargs>=0.5.0andrequires-python >=3.7;uv runcannot resolve the project because that range includes Python 3.7.0-3.7.3 whiledargsdepends ontypeguard>=4, which requires at least Python 3.7.4. Conda metadata still advertises Python >=3.6 and omits runtime dependencies present inpyproject.toml.Relevant code
dpdispatcher/pyproject.toml
Lines 22 to 31 in 98a9e08
dpdispatcher/conda/meta.yaml
Lines 22 to 35 in 98a9e08
dpdispatcher/.github/workflows/test.yml
Lines 20 to 25 in 98a9e08
Observed locally
uv run --no-dev python -c 'import sys; print(sys.version)'fails with an unsatisfiable dependency split forpython_full_version >= '3.7' and < '3.7.4'.Impact
New development environments can fail to resolve. Conda packages can miss imports such as
pyyamlandtomlion older Python versions. CI also skips Python 3.8 while the package classifier declares support.Suggested fix
Align the supported Python floor and CI matrix with dependency reality, for example
requires-python >=3.7.4or a higher version if intended. Update conda runtime dependencies to includepyyamland conditionaltomli, and align conda's Python requirement with PyPI metadata.