py::print doesn't acquire the GIL ?

[HunterBelanger]

I have a library which uses spdlog for the output, and to make sure things work properly with Jupyter notebooks, I use custom made sinks which just call py::print to make sure everything goes through the Python stdout. In modifying this library to work with threading in Python, I noticed that when multiple threads try writing to the log, I get GIL errors, because multiple threads are trying to output at the same time. If I add a py::gill_scoped_acquire right before I call py::print, this problem does not occur (which makes sense to me).

I was just surprised by this behavior though, as based on the GIL documentation, I would have expected py::print to acquire the GIL automatically. It seems to me that py::detail::print should have a gil_scoped_acquire to prevent this issue, as from what I can tell, it will always be accessing a Python object when writing. Is there a reason to not have this here that I am missing ? If this is something that should be there, I am happy to add it myself; I just want to make sure there isn't a reason to not have it, or there isn't a better way to accomplish this.