1

Let's assume we have the following dependency tree in Python using PIP as package installer:

A==1.2.1:
|  - B==1.5.4
|  - C==?.?.?

D==1.3.0:
|  - C==?.?.?

Let's also assume that the existing versions of package C are the followings:

- 0.0.8
- 1.0.2

As we know, PIP will not install different versions of the same package into an environment. Which means it has to pick one suitable version for all dependents.

What I simply cannot understand how PIP determines the suitable version.

If package A depends on C==0.0.8 and package D depends on C==1.0.2, this situation cannot get easily handled, unlike other package managers such as NPM could do.

Johnny
  • 337
  • 3
  • 15
  • 1
    In that case `pip` should report an error, as it cannot satisfy both dependencies at the same time. Did you check this with some example projects? – a_guest Jul 01 '20 at 14:53
  • I wonder how does `npm` resolves contradicting dependencies? I think it's simply not possible. Or maybe `npm` installs multiple versions of a dependency to be used by different consumers? – phd Jul 01 '20 at 16:27

2 Answers2

1

When pip installs a package, it automatically installs any dependent Python packages without checking if these conflict with previously installed packages. It will install a package and any of its dependencies regardless of the state of the existing installation. Because of this, a user with a working installation of, for example, Google Tensorflow, can find that it stops working having used pip to install a different package that requires a different version of the dependent NumPy library than the one used by Tensorflow. In some cases, the package may appear to work but produce different results in detail.

Anaconda (Python distribution) - Wikipedia

Johnny
  • 337
  • 3
  • 15
1

On this topic, things will change relatively soon. Since pip's developers are currently working on a new dependency resolver:

To test it today, you might want to install pip 20.2b1 and enable the unstable feature "resolver", with for example either one of the following:

  • PIP_UNSTABLE_FEATURE=resolver python -m pip install SomeProject
  • python -m pip --unstable-feature=resolver install SomeProject

See the following link for more details:

Other references:

Update pip 20.2

sinoroc
  • 18,409
  • 2
  • 39
  • 70