M-NO-GLOB-REEXPORTS is excessively broad

[sandersaares]

It is perfectly fine to re-export types within a crate for filesystem-structural reasons, if we want to split some types off to a separate file. Glob re-exports should only be discouraged for cases where the re-exported item is already public via a different path (e.g. part of a different crate or via another public module).

For filesystem-structural reasons, no review of what is being re-exported is expected, either - each file should have clear pub or lack thereof on its types, indicating what is part of the public API. As an alternative UX, generated documentation can be used for review.

[ralfbiedert-microsoft]

It is perfectly fine to re-export types within a crate for filesystem-structural reasons, if we want to split some types off to a separate file.

The rule does not prevent you from doing that, it only asks you to name the re-exported types.

Glob re-exports should only be discouraged for cases where the re-exported item is already public via a different path (e.g. part of a different crate or via another public module).

Can you give an example? This seems to be unrelated to the issue of the glob.

For filesystem-structural reasons, no review of what is being re-exported is expected, either

It's not to review the items being re-exported, it's about being able to review what the current module actually re-exports.

[sandersaares]

it's about being able to review what the current module actually re-exports.

I will try phrase my criticism in this context: this is not pragmatic.

There already exist ways to get this information, such as opening the API documentation or the target file being re-exported from and looking for its pub types. In its current form, disabling glob re-exports adds burdensome manual labor to maintaining a list of types that under this interpretation exists only for review purposes. The reason I mention filesystem-structural re-exporting is that this requires a lot of re-exporting, thereby increasing the burden from "annoying" to "unacceptable" in my eyes.