Ben Thompson

Yeah. It's an awesome idea, but it loses a lot of the potential of garbage collection to improve performance because it uses raw pointers. That means that first, it can't...

Thanks! Good suggestions. I'll work on this at some point soon.

Would it be feasible to do something like this: ``` version_str = '1.0.1' with open('version', 'w') as f: f.write(version_str) cfg['dependencies'].append('version') ``` The [`'dependencies'` configuration variable](https://github.com/tbenthompson/cppimport#does-cppimport-recompile-every-time-a-module-is-imported) doesn't actually do anything with...

The snippet that I wrote above would work with no changes. In practice, I'd probably write out the `version` file as part of the process that updates dependencies. Or, do...

Here's a more explicit version of what I'm imagining. Two files: ``` // cppimport - my_module.cpp // you c++ code here ... /* */ ``` ``` # python_file.py import cppimport...

Yes, it's as intended that the template is not being run a second time here. You can see the check for whether to enter the build step here: https://github.com/tbenthompson/cppimport/blob/e19e9a601bd1deeae983ef02aa3aa78171472d3f/cppimport/__init__.py#L86 Currently,...

WIP: https://github.com/tbenthompson/cppimport/pull/80

no time to continue this now, but I'll get back to this the next time I need to push a release.

There's also a bump version workflow that I could manually trigger to start the process: https://github.com/MylesBorins/node-osc/blob/main/.github/workflows/bump-version.yml

Sorry for such a delayed response! I would be supportive of a change like this. I think your proposed API is pretty reasonable. I won't have time to work on...