What are our options for installing the most recent dev version?

[ryx2]

Question

My team is excited about the toolsets functionality in Agent so then we can much more easily integrate our mcp's with our tools #2024. The only issue is this hasn't been released yet on 0.4.3. In order to build with it, we'd like to install our pydantic with the github main branch, but are unsure how since there seems to be some sort of dependency issue described below.

Is there a standard way we can pull from head to develop off of there if there a future dev features that we like that we want to integrate?

❯ poetry add git+https://github.com/pydantic/pydantic-ai.git


Updating dependencies
Resolving dependencies... (0.0s)

Because pydantic-ai (0.4.4.dev19+01c550c2) @ git+https://github.com/pydantic/pydantic-ai.git@main depends on pydantic-ai-slim[ag-ui,anthropic,bedrock,cli,cohere,evals,google,groq,huggingface,mcp,mistral,openai,vertexai] (0.4.5.dev19+01c550c2) which doesn't match any versions, pydantic-ai is forbidden.
So, because 0phones depends on pydantic-ai (0.4.4.dev19+01c550c2) @ git+https://github.com/pydantic/pydantic-ai.git@main, version solving failed.

Additional Context

No response

[xjose97x]

@ryx2 the following command should install the latest version of pydantic-ai (the latest commit from the main branch):

uv add "pydantic-ai @ git+https://github.com/pydantic/pydantic-ai.git" --branch main

[ryx2]

^ why would that be different than poetry adding it?

[DouweM]

@ryx2 The pydantic-ai top-level package depends on the pydantic-ai-slim package defined in the same repo, and automatically installs it with all optional dependency groups. poetry doesn't seem to realize that pydantic-ai-slim should be installed from the same repo, so it's complaining that the requested version doesn't exist. uv does know it's inside the same repo because of this:

pydantic-ai/pyproject.toml

Lines 67 to 72 in 87871b3

	[tool.uv.sources]
	pydantic-ai = { workspace = true }
	pydantic-ai-slim = { workspace = true }
	pydantic-evals = { workspace = true }
	pydantic-graph = { workspace = true }
	pydantic-ai-examples = { workspace = true }

We strongly recommend using uv rather than poetry!