Skip to content

[Python Integration] CFFI does not fail properly - leaving Fortran running #457

@FlorianDeconinck

Description

@FlorianDeconinck

Our integration - being the one in GEOS Physics via MAPL or the custom GEOS FV code - is no longer properly crashing when an error is raise in Python, unless that error is a system raise by Python itself.

The original feature was to caught the raise, forward it to C++ and properly exit(-1) there. It no longer works for reasons unknown.

Our colleagues at CSCS/MeteoSwiss seems to run into the same issue -> C2SM/icon4py#1245

DoD:

  • Create a trap that fails integrated in Fortran on user python raise
  • Create a trap that fails integrated in Fortran on stencil/orchestration compilation error
  • [Optional] Make a unit test

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels
    No fields configured for Feature.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions