New upgrade-strategy eager-while-respecting-dependencies

[bersbersbers]

What's the problem this feature will solve?
Image you are working with the newest version of package A that depends on B==2. You have B==1 installed, and the most recent version is B==3.

How do you upgrade to the newest allowed version of B, B==2, without going through all dependencies of A manually? If I am not mistaken,
pip install A --upgrade [--upgrade-strategy only-if-needed] keeps B==1
pip install A --upgrade --upgrade-strategy eager updates to B==3
pip-review --interactive will (offer to) upgrade to B==3

A concrete example:
pip install PyInstaller --upgrade --upgrade-strategy eager will upgrade to setuptools==40.4.1, even though tensorflow-gpu==1.11.0rc1 is installed, which requires setuptools<=39.1.0.

Describe the solution you'd like
pip install A --upgrade --upgrade-strategy eager-while-respecting-dependencies could solve that, parsing the dependency tree for the maximum allowed version of B and upgrading to that one.

[RonnyPfannschmidt]

sounds like a duplicate of #988

[bersbersbers]

@RonnyPfannschmidt I agree, yes. My specific question about a new upgrade-strategy is a bit more specific than that huge discussion over there, but it captures the essence of my problem.

[RonnyPfannschmidt]

@bersbersbers my understanding is that solving #988 would sort out the issue you see with eager updates correctly, and making that particular update strategy actually work would require it to begin with

[bersbersbers]

Agreed!

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.