experimental/sources/: resolvelib-1.0.1 metadata and description

Homepage Simple index

Resolve abstract dependencies into concrete ones

author Tzu-ping Chung
author_email uranusjr@gmail.com
  • Development Status :: 3 - Alpha
  • Intended Audience :: Developers
  • License :: OSI Approved :: ISC License (ISCL)
  • Operating System :: OS Independent
  • Programming Language :: Python :: 2
  • Programming Language :: Python :: 3
  • Topic :: Software Development :: Libraries :: Python Modules
description_content_type text/x-rst
keywords dependency,resolution
license ISC License
provides_extras release
File Tox results History
21 KB

ResolveLib at the highest level provides a Resolver class that includes dependency resolution logic. You give it some things, and a little information on how it should interact with them, and it will spit out a resolution result.

Intended Usage

import resolvelib

# Things I want to resolve.
requirements = [...]

# Implement logic so the resolver understands the requirement format.
class MyProvider:

provider = MyProvider()
reporter = resolvelib.BaseReporter()

# Create the (reusable) resolver.
resolver = resolvelib.Resolver(provider, reporter)

# Kick off the resolution process, and get the final result.
result = resolver.resolve(requirements)

The provider interface is specified in resolvelib.providers. You don’t need to inherit anything, however, only need to implement the right methods.


The intention of this section is to unify the terms we use when talking about this code base, and packaging in general, to avoid confusion. Class and variable names in the code base should try to stick to terms defined here.

Things passed into Resolver.resolve() and provided by the provider are all considered opaque. They don’t need to adhere to this set of terminologies. Nothing can go wrong as long as the provider implementers can keep their heads straight.


A thing that can be installed. A Package can have one or more versions available for installation.


A string, usually in a number form, describing a snapshot of a Package. This number should increase when a Package posts a new snapshot, i.e a higher number means a more up-to-date snapshot.


A collection of one or more Versions. This could be a wildcard, indicating that any Version is acceptable.


A combination of a Package and a Version, i.e. a “concrete requirement”. Python people sometimes call this a “locked” or “pinned” dependency. Both of “requirement” and “dependency”, however, SHOULD NOT be used when describing a Candidate, to avoid confusion.

Some resolver architectures refer this as a “specification”, but it is not used here to avoid confusion with a Specifier.


An intention to acquire a needed package, i.e. an “abstract requirement”. A “dependency”, if not clarified otherwise, also refers to this concept.

A Requirement should specify two things: a Package, and a Specifier.


Please see developer documentation.