<@U03AV0GG20H> nice progress on that mega PR!
# contribute
i
@Stefan Krawczyk nice progress on that mega PR!
s
still some more validation to go
but it's looking okay.
i
🥳
s
for the MCP one -- how do we want to think about the python dependencies on pandas and numpy in it?
i
Please elaborate
s
it imports them -- should they be a dependency in the mcp pyproject.toml target? right now we assume it's there because of hamilton - -just thinking about the future
i
Hmm there are plans to remove pandas and numpy as dependencies, right? So according to the same logic - those should be optional deps. If the user doesn't need pandas no reason to install it imho
The functionality should be conditional - if import succeeds or not
s
yep
i
I addressed your concern by adding a "capability" (optional deps) discovery tool/stage, as well as instructions on how to start the mcp with the desired dataframe library