ResolutionImpossible error message does not distinguish between extras

[ZanSara]

Description

If a package has extras with conflicting dependencies, the error message displayed by pip does not clarify which extras triggered the conflict, and also produces an error message with is rather confusing. Example (see steps to reproduce):

The conflict is caused by:
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas<2 and >=1.3
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.1

This happens when extra1 asks for pandas==1.4.0 and extra2 asks for pandas==1.4.1. extra3 has nothing to do with the conflict, but from the error message it's impossible to tell. In addition, the error message never mentions neither which extras that caused the issue, nor the actual issue, as pandas==1.4.0 is never mentioned.

Expected behavior

It should point to the extras that triggered the ResolutionImpossible error in a clear way.

pip version

22.0.2

Python version

3.9

OS

Ubuntu 21.10

How to Reproduce

Create a setup.cfg with the following content:

[metadata]
name = resolution_impossible_with_extras

[options]
install_requires =
    pandas >=1.3, <2

[options.extras_require]
extra1 = 
    pandas==1.4.0
extra2 = 
    pandas==1.4.1
extra3 = 
    numpy==1.22.1

and an "empty" setup.py like the following content:

from setuptools import setup
setup()

Then execute in the same folder:

pip install .[extra1,extra2,extra3]

Output

Processing /path/to/folder/resolution-impossible-with-extras/package
  Preparing metadata (setup.py) ... done
Requirement already satisfied: pandas<2,>=1.3 in /path/to/folder/resolution-impossible-with-extras/venv/lib/python3.9/site-packages (from resolution-impossible-with-extras==0.0.0) (1.4.0)
ERROR: Cannot install resolution-impossible-with-extras[extra1,extra2]==0.0.0 because these package versions have conflicting dependencies.

The conflict is caused by:
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas<2 and >=1.3
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.1

To fix this you could try to:
1. loosen the range of package versions you've specified
2. remove package versions to allow pip attempt to solve the dependency conflict

ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts

Code of Conduct

• I agree to follow the PSF Code of Conduct.

[q0w]

Sometimes, pip provides the different output:

The conflict is caused by:
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas<2 and >=1.3
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.0
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.1

Its inconsistent, because Criterion object in failure_causes from resolvelib can contain 2 or 3 SpecifierRequirement objects

[ZanSara]

I haven't managed to get pip to output the message you show here. How did you obtain it?

[q0w]

Run it several times.

[q0w]

I also checked pip v21 and it also produced different outputs.

[notatallshaw]

I also checked pip v21 and it also produced different outputs.

Seems this different output has been noticed in a couple of other issues #10824 and #10391 (comment)

[q0w]

logs:
1.txt
2.txt

[uranusjr]

If anyone’s interested in improving this, see also #9237.

[notatallshaw]

FYI the error message does now display the extra being refereed to, if you run on latest Pip you get:

The conflict is caused by:
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas<2 and >=1.3
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.0; extra == "extra1"
    resolution-impossible-with-extras[extra1,extra2,extra3] 0.0.0 depends on pandas==1.4.1; extra == "extra2"

[notatallshaw]

Closing as this now appears to be fixed and there was no further feedback. Feel free to open a new issue if there is a new way of reproducing the problem.