Skip to content

Conversation

@BenjaminRodenberg
Copy link
Contributor

@BenjaminRodenberg BenjaminRodenberg commented Aug 12, 2025

The goal of this PR is to prepare the branch yac-coupling for the integration into devel:

  • I integrated the mpi_handshake feature (needed by YAC and XIOS) with the colouring scheme in order to avoid conflicts between the two schemes and avoid a degradation.
  • I added additional checks to make sure the provided dependencies YAC and/or XIOS are compatible and Elmer is initialized correctly.

Note: Currently, a Windows test is failing. I'm investigating and hopefully fixing this issue in #691

@BenjaminRodenberg BenjaminRodenberg changed the title WIP: Refactoring and fix Windows WIP: Refactoring of colouring and fix Windows Aug 12, 2025
@BenjaminRodenberg BenjaminRodenberg force-pushed the yac-coupling-refactoring branch from 70a269a to 370d5a7 Compare August 26, 2025 07:30
@BenjaminRodenberg
Copy link
Contributor Author

Note that the Windows tests are still failing. I would suggest to move this topic to another PR if it is a larger issue.

@BenjaminRodenberg BenjaminRodenberg force-pushed the yac-coupling-refactoring branch from 4a4a732 to c71e151 Compare August 27, 2025 10:31
@BenjaminRodenberg BenjaminRodenberg changed the title WIP: Refactoring of colouring and fix Windows WIP: YAC-coupling integration in Elmer (colouring and mpi_handshake) Aug 27, 2025
@raback
Copy link
Contributor

raback commented Sep 3, 2025

The Windows tests seem to be failing but does not seem to be too severe. The first one fails tests "Contact" and "Shell_with_Solid_Beam_Eigen". Upon running again both pass. The 2nd one with external dependencies only fails "Contact" and passes that on rerun. The "Contact" test has shown this kind of instability also before. From the output it would seem that for some reason the "TEST_PASSED" file is not available before its inspection. For some reason it only happens to this test. I will have a look what could make that case so special....

@BenjaminRodenberg
Copy link
Contributor Author

Thank you for looking into this @raback ! So I understand we do not have to worry about the Windows tests here and can proceed?

What's still missing from my perspective in this PR are a few (manual and maybe even automated) tests. I would like to make sure this change does not introduce a regression if XIOS is used without YAC.

Next steps:

@BenjaminRodenberg BenjaminRodenberg force-pushed the yac-coupling-refactoring branch 2 times, most recently from 64169bc to 88332e1 Compare October 14, 2025 13:58
Copy link
Contributor Author

@BenjaminRodenberg BenjaminRodenberg left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm a bit unsure about whether we agreed to keep elmer_grid.{c,h} or not. Rest is from my perspective good to go.

Still needs some manual testing. I would suggest to add automated testing at a later point in time and get this feature merged.

* Avoids (potential) bug if XIOS is used without handshake
* Not necessary for YAC but still easier to understand this way
* Use PARAMETER for defining config file name at beginning.
* Factor checking for config files out since this is less error prone if
  features are nested.
* Use PARAMETER for defining config file name at beginning.
* Factor checking for config files out since this is less error prone if
  features are nested.
@BenjaminRodenberg BenjaminRodenberg marked this pull request as draft November 5, 2025 10:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants