There is very little benefit for that as pytest doesnt use load_entry_point or pkg_resources.resolve , so the installed egg-info constraints are not in play at all, even when they do become invalid.
Compare that with the approx hundred d-l-py* currently invalid egg-info constraints which are currently causing problems because they contain an executable entry point, which as you should have realised I have been actively fixing.
Anyway, obviously the hazing continues, so I'll do it.
See what I did for the constrains in pytest-shutil or pytest-virtualenv from this guy:
python-pytest-shutil/python-pytest-shutil.spec:sed -i -e 's:pytest<4.0.0:pytest:g' setup.py python-pytest-virtualenv/python-pytest-virtualenv.spec:sed -i -e 's:pytest<4.0.0:pytest:g' setup.py
I even tested those to see they work with pytest4...
There is very little benefit for that as pytest doesnt use load_entry_point or pkg_resources.resolve , so the installed egg-info constraints are not in play at all, even when they do become invalid.
Compare that with the approx hundred d-l-py* currently invalid egg-info constraints which are currently causing problems because they contain an executable entry point, which as you should have realised I have been actively fixing.
Anyway, obviously the hazing continues, so I'll do it.