-
Notifications
You must be signed in to change notification settings - Fork 33
Packaging improvements #9
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
We just discovered on the MCP SDK Quickstart documentation that is indeed about starting with a |
Dear amotl Thx for the PRs! Yeah the project is getting quite the attention, so it makes sense to push it to pypi. EDIT: Ah I've never actually tested the pip installation instructions, those were from a PR, sry about that. I'll revert the documentation to use the manual git clone method for now, and create a proper package in a unique PYPI namespace early next week. Thx, a lot for bringing this to my attention! |
… lack of proper namespacing. See runekaagaard#9" This reverts commit b85aae6.
Hi Rune, the installation works perfectly well now, and will no longer collide with others that use With kind regards, NB: We are validating your package over here against CrateDB, which is compatible with PostgreSQL, but provides a dedicated dialect. That also spawned a relevant improvement that was needed to make it work. 🍀 |
Dear Rune,
that's an excellent software package, thanks a stack for conceiving it. It works from the start, modulo a few obstacles we observed when installing it.
Would you accept a patch that improves packaging a bit more, by adding a few touches that whip the modules into the
mcp_alchemy
namespace, and expandpyproject.toml
to make it a real package that could also be uploaded to PyPI? 1With kind regards,
Andreas.
Footnotes
Currently, while installation per
pip install ...
works, the package will install a module calledserver
into the top-level Python library tree at hand, which is unfortunate, because it can easily yield collisions. This is not a hypothetic thing, becausepg-mcp
actually does the same, so they can't be installed together into the same Python environment. 2 ↩We hope this specific detail is not a flaw on the recommendations/conventions of MCP itself, how to build an MCP server? If so, relevant documentation should dearly be fixed. If you know anything about it, please let us know. Otherwise, I think it was just an accidental collision that can easily happen when software is still in its infancy. NB: We submitted the same message also to @stuzero per https://github.com/stuzero/pg-mcp/issues/10. ↩
The text was updated successfully, but these errors were encountered: