This seems like a deliberate choice, but it seems that at the moment the Author: metadata is pulled preferentially from maintainer=, and Maintainer: is never actually generated.

The implementation is here, and the implementation of get_contact that prioritizes maintainer is here.

I suspect that this is an artifact of the fact that PEP 314 does not provide a Maintainer field, and that was a later addition in PEP 345?

I can make a PR for this, but there is an implementation question. The options I see right now are:

1. Maintain the current behavior and simply add a "Maintainer" field. This is the most "backwards compatible" way to do things.
2. Change the behavior so that author= is given preference over maintainer= when generating the AUTHOR field, but still use maintainer if no author is specified. Additionally always add the Maintainer field if maintainer= is present.
3. Change the behavior so that only author= generates the Author field and only maintainer= generates the Maintainer: field. This is the least backwards-incompatible change.

I suspect that the choice of these probably should come down to how PyPI and Warehouse handle the metadata. If both PyPI and Warehouse are able to handle an either/or Author/Maintainer field, then I suggest that the best option is 3. Basically the fallback logic should be on the metadata consumer side, not on the metadata producer side.