Daniele Nicolodi

If we use a stable build directory, I am not sure about what should happen when a package is build twice with different configurations (either meson arguments passed via the...

Given a build directory, the location of the meson logs is know, thus there is no need to parse the build output to get it. If a build directory is...

One issue is that the detailed setup log can be many thousands of line long and is most often not necessary to debug `meson setup` failures. On the contrary, showing...

Another aspect to consider is that having a `print-log` option requires running the setup stage one more time to get the log after the user encounters a failure. While having...

In Github Actions, there is also a way to tag output in a way that it is hidden by default, and you need to click on the small arrow to...

@rgommers This is now working. We would just need to plug in a function to compute `ncores` and some documentation.

It seems that SciPy uses only the physical cores, thus I'll be temped to implement only that for now, and implement logical core detection only if someone asks for it....

Can we do that from a licensing point of view? `meson-python` is released under the MIT license, `loky` and SciPy are licensed under the BSD-3 license.

I was thinking to do just that, I just wanted to make sure that it is ok to include some BSD code in `meson-python`.

Do you have a preference for `%(ncores)d` vs `{ncores}` vs `${ncores}` namely old-style `%` string formatting, vs new-style string formatting, vs template strings? The only practical difference I can see...