Declarative modules: submodule declarations are not considered covered

[alex]

From the coverage report on pyca/cryptography#11159

As you can see the mod ... lines are all considered uncovered. This possibly has something to do with the spans on the generated code.

[alex]

I suspect, but am not positive, that this is because the __pyo3_init functions are generated for submodules, but obviously never called.

[alex]

Ok, my idea here is:

• Add a new submodule parameter to pymodule: pymodule(submodule)
• Automatically add this attribute for nested modules (users could add it themselves for their own modules)
• When a module is marked submodule, we no longer emit __pyo3_init

WDYT?

[alex]

@davidhewitt Once this and #4288 and #4308 are merged, I'd be interested in doing a point release so we can take advantage of declarative modules. Anything else you'd want to get in there?

[davidhewitt]

I just had #4304, agreed a point release would be good. I'll try to review / prep the release tomorrow at the latest.

[alex]

Cool, thanks!

[davidhewitt]

I assume this is now resolved 😄

[alex]

Yup! We're all aboard adopting them: pyca/cryptography#11158

…

On Tue, Jul 9, 2024 at 8:16 AM David Hewitt ***@***.***> wrote: I assume this is now resolved 😄 — Reply to this email directly, view it on GitHub <#4286 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAAGBDTCZRXOSN2QCHGUR3ZLPIDBAVCNFSM6AAAAABJ4X44VSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMJXGQ4TENJWGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- All that is necessary for evil to succeed is for good people to do nothing.