Dependency Conflict: libp2p==0.4.0 requires trio>=0.26.0, which causes incompatibility with modern pytest-trio

[p4ngui]

Summary

Hello py-libp2p team,

I've encountered a significant dependency conflict when using libp2p==0.4.0 with modern versions of
pytest-trio. The core of the issue is a three-way incompatibility between libp2p, trio, and pytest-trio.

Expected behavior

Would it be possible to release a new version of py-libp2p (e.g., 0.4.1) that updates the host.run()
method to be compatible with the modern trio and pytest-trio APIs? Specifically, making it gracefully
handle the task_status keyword argument would resolve this dependency hell and make it much easier to
integrate py-libp2p into projects with up-to-date testing frameworks.

Thank you for your time and for maintaining this great library!

Actual behavior

Problem:

1. libp2p==0.4.0 has a dependency on trio>=0.26.0.
2. trio>=0.26.0 has removed the trio.MultiError exception, which was a key part of its older API.
3. Modern versions of pytest-trio (e.g., >=0.8.0) are designed to work with these newer versions of
   trio.
4. However, the way libp2p==0.4.0's host.run() method interacts with the modern pytest-trio test runner
   leads to a TypeError: BasicHost.run() got an unexpected keyword argument 'task_status'. This happens
   because pytest-trio injects a task_status argument that the older libp2p run method doesn't accept.

What I've Tried:

• Using older pytest-trio: I tried downgrading to pytest-trio==0.7.0 to avoid the task_status issue.
  This failed because pytest-trio==0.7.0 requires an older version of trio that still has
  trio.MultiError, which conflicts with libp2p's requirement of trio>=0.26.0.
• Code Workarounds: I attempted various workarounds in my application code (using lambda,
  functools.partial, and explicit wrapper functions) to intercept the task_status argument, but none of
  these were successful, which suggests the incompatibility runs deeper than just argument passing.

Relevant log output

Possible Solution

update trio library

Environment

Environment:
   * python: 3.12
   * libp2p: 0.4.0
   * trio: >=0.26.0
   * pytest-trio: >=0.8.0

Would you like to work on fixing this bug ?

No

[seetadev]

@p4ngui : Thank you for the clear report and for documenting the dependency conflict so well — this is extremely helpful for us. 🙌

You're absolutely right: the interaction between trio ≥ 0.26, the removal of MultiError, and the newer pytest-trio task lifecycle (especially the injected task_status argument) exposes gaps in our current host.run() implementation. This is a valid issue and something we should address to keep py-libp2p compatible with modern async testing workflows.

We’ll take this forward on our side.

I’m looping in @sumanjeet0012 and @acul71, who will work together with me and @pacrob on a fix, and propose an update to ensure host.run() gracefully handles task_status in line with recent Trio APIs.

We’ll update the issue soon with a concrete patch plan and a timeline for a 0.4.1 bugfix release.

Thanks again for raising this and for using py-libp2p — we really appreciate it! 🚀

[acul71]

@p4ngui This issue has been fixed by #1074
See if you dig it