importlib_metadata
importlib_metadata copied to clipboard
python
Backport of the importlib.metadata module
Python 3.9 is the lowest version supported here, we can make some little cleanups.
`napari` uses the entry-point value for non-object references, i.e. a file path to load metadata from (i.e. as `module:file_path`). This pattern seemed like a good idea for certain use cases....
In use with multiprocessing (e.g. when pickling `awkward` arrays to transmit them from a subprocess to the host), I see issues with `importlib_metadata`. This was originally observed by @richeldichel as...
This is a very similar issue to https://github.com/python/importlib_metadata/issues/392, but for me it is happening on py3.8. #### Steps to reproduce (on py3.8) `pip install --no-binary=:all: importlib-metadata` #### Relevant output log...
On the [pypi.org page](https://pypi.org/project/importlib-metadata/), it says under license "Apache Software License". Since this project uses MIT this should probably be updated.
Before, one could get OSError 22 and BadZipFile errors due to re-used file pointers in forked subprocesses. This implements the fix recommended by @anntzer . Fixes #520
> I've also observed that the issue affects `.module` and `.attr` and `.load`. I'm trying to decide what's the best behavior here. I think it should fail early, at construction...
