Skip to content

Develop a way to figure out the maximal list of projects that _may_ be compatible with WASI #42

Open

Description

We are after a list of potential projects that could work with Python under WASI. There's essentially 6 layers to this, each one more specific than the previous, but all accurate for what they claim:

  1. All projects on PyPI
  2. Projects on PyPI that have a pure Python wheel
  3. Projects who only have dependencies that are pure Python wheels
  4. Importation of the top-level project succeeds
  5. Test suite passes
  6. Projects declare their support

We can calculate layers 1 and 2 easily thanks to https://github.com/sethmlarson/pypi-data (although with slightly outdated data). We can calculate layer 2 with sqlite3 pypi.db 'SELECT DISTINCT package_name FROM wheels WHERE python = "py3" OR python = "py2.py3" ORDER BY package_ name ASC' > project_with_pure_wheels.txt.

Layer 3 requires writing a resolver which can resolve for a specific environment and only for specific wheel tags (which helps out with lock files, but it's still work). Layer 4 probably can be automated with introspection inside wheel files and importing module(s) based on what's found in them (luckily WASI is sandboxed, so this should actually be a safe operation). Layer 5 will either have to be some crazy set of heuristics or we accept it as being a per-project thing (plus being able to actually run pytest under WASI). And for layer 6, I have opened https://discuss.python.org/t/do-we-want-classifiers-for-webassembly-on-pypi/22712 to discuss it with the community.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions