Conversation
|
@RobPasMue @dipinknair @SMoraisAnsys I can't reproduce the CI failures locally and it seems the upgrade to 3.13 didn't have any CI issue. Any ideas? |
|
Looks like your implementation is getting "activated" somehow even if the user doesn't opt-in? https://github.com/ansys/ansys-pythonnet/actions/runs/12750835469/job/35536486536?pr=23#step:11:75 The failing tests talk about internal properties being accessed. The tests are expected to raise an error because the user shouldn't be able to access them but now, with your implementation, they are accessible by default and do not raise an error. |
|
I'll try to reproduce that locally and come back once I'm done :) |
I'm able to reproduce the CI failure locally and passed main locally. As @RobPasMue mentioned, the implementation is at cause here. |
|
@RobPasMue @SMoraisAnsys @dipinknair this is ready for review |
RobPasMue
left a comment
There was a problem hiding this comment.
Fine by me - should we consider bringing these changes upstream? In any case, I would include in the main README of our fork some info regarding this implementation
|
@RobPasMue I started to propose the binding manager upstream first. If that is accepted, I can propose the explicit interface implementation option too. Yes - the readme should be updated. Should we do that on pypi releases or on merges to main? |
Awesome, thanks for handling that!
I would do it as part of this PR. Our main branch should reflect the change already. We can do a follow up release right after. |
Fixes #22
Gives the user an option to opt-into exposing explicit interface implementations.
This is done by adding a place to register customizations for python bindings by type or assembly.
Usage (to customize binding for all types in an assembly):