There is a mismatch on the cffi requirement between the conda and pip recipes. Is this intentional and if not, can they be updated to match? As a result, my rattler packaging fails on a pip check as the dependencies can't resolve. I have other packages I'm using that require cffi < 2. I would prefer having to manually pin pynacl < 1.6.1 and if the current version was marked to require cffi >= 2 that would fix the issue for me..
The pip package requirements are as follows:
[build-system]
# Must be kept in sync with `dependencies` below
requires = [
"setuptools>=61.0.0,!=74.0.0",
"wheel",
"cffi>=1.4.1; platform_python_implementation != 'PyPy' and python_version < '3.9'",
"cffi>=2.0.0; platform_python_implementation != 'PyPy' and python_version >= '3.9'",
]
Conda requirements are as follows:
host:
- _libgcc_mutex 0.1 conda_forge
- _openmp_mutex 4.5 2_gnu
- bzip2 1.0.8 hda65f42_8
- ca-certificates 2025.11.12 hbd8a1cb_0
- cffi 2.0.0 py311h03d9500_1
...
run:
- __glibc >=2.17,<3.0.a0
- cffi >=1.4.1
- libgcc >=14
- libsodium >=1.0.20,<1.0.21.0a0
...
There is a mismatch on the cffi requirement between the conda and pip recipes. Is this intentional and if not, can they be updated to match? As a result, my rattler packaging fails on a pip check as the dependencies can't resolve. I have other packages I'm using that require
cffi< 2. I would prefer having to manually pin pynacl < 1.6.1 and if the current version was marked to requirecffi >= 2that would fix the issue for me..The pip package requirements are as follows:
Conda requirements are as follows: