Incase you wanted to build on MacOS

[mikeohyan]

In case you wanted MacOS build instructions:

What you need before hand:

• Install XCode from the Apple App Store (this should get clang setup)
• uv to manage python environments (https://docs.astral.sh/uv/getting-started/installation/)

To confirm you can compile some C++ code, in a terminal you should be able to run: clang --version

Now to compile:

• git clone https://github.com/robotpy/mostrobotpy.git to a working directory
• in that directory, run uv init to setup python package management
• install the package requirements: uv add -r rdev_requirements.txt
• install numpy uv add numpy
• uv add pip

Now to run compile command, instead of the bash script, we're going to run this instead.
In the working directory, run:
uv run python -m devtolls ci run

This will run for... a while, if this works you'll get a bunch of wheels (.whl) in <working dir>/dist

Now to use these wheels, in your python project (assuming you are using uv to manage packages):
Run uv pip install <whl path> for whatever you need.
There are packages that will have dependencies, if asked, install what's missing first.
Some packages are ok with the modifer: uv pip install --no-deps <whl path>

--
NB: Jumped through these steps to get NetworkTables working for a project.

[virtuald]

We distribute wheels for macOS, is there a reason they don't work for you? Or are you just providing this because the instructions don't currently exist?

[mikeohyan]

We thought you did! When trying to do a standard install (uv add ...), it didn't seem to be able to solve for MacOS. I think there was a wheel dependency that wasn't solving correctly for the 2025.* versions.

But also a nice to have, in case someone wanted to know where a wheel came from.

[virtuald]

You should be able to do a uv pip install pyntcore to get ntcore support for macOS, there is a wheel published at https://pypi.org/project/pyntcore/#files

It is possible that our dependencies are weird and the uv project support might not be functioning, I haven't used uv very much.

We probably should document the build process better, maybe in build.md in the root directory.

[auscompgeek]

My team uses uv for our robot code, and everything resolves correctly. I know because I tested it on both Linux and macOS 😄

[TheTripleV]

I do understand that it's a bit confusing considering on pypi

• ntcore is unrelated
• networktables doesn't exist
• pynetworktables is abandonware but nowhere does it point to pyntcore
• pynetworktables2js is abandonware

[mikeohyan]

You should be able to do a uv pip install pyntcore to get ntcore support for macOS, there is a wheel published at https://pypi.org/project/pyntcore/#files

It is possible that our dependencies are weird and the uv project support might not be functioning, I haven't used uv very much.

We probably should document the build process better, maybe in build.md in the root directory.

Just digging into uv for a bunch of new builds right now. It's the first manager that feels pretty good for python.

Happy to support, glad super responsive ppl are working on this. GL w/ your bots!