Andy Porter

I've been bold and added a new `inline_calls` method to `examples/nemo/scripts/utils.py` that attempts to inline everything. I've added a call to this from the `acc_kernels_trans.py` script so we'll see what...

As usual, actually trying things for real has revealed a couple of problems - one to do with generic interfaces and another with the `psyclone_internal_cmp` routine that we introduce in...

Integration tests failed but I can't reproduce it interactively. Therefore, I've fired them off again but requested that the failing build be done in serial to help me see what's...

Issue was that a local routine that we were trying to inline was in a CodeBlock and thus not found. I was raising an InternalError in this case (because I...

Finally have integration tests working again. Looking at test coverage I now realise that I have duplication between `InlineTrans._find_routine()` and the new (in this PR) `Call.get_callees()`.

Most of the duplication is gone now. I've started trying to integrate with the `ModuleManager` but have hit the issue that it's currently LFRic specific. I confess that I reviewed...

Integration tests failed as it turns out that the Fortran frontend wasn't passing the include paths when creating a new parser instance. If I fix that then processing a real...

It turns out that the OpenMPI header files include quite a few `EXTERNAL` statements. If the header is INCLUDEd at the top level of a MODULE this results in the...

I've fired off the integration tests for this. Note that it currently still includes the regex workaround/extension to the ModuleManager that I think @hiker was going to revisit?

Integration tests (processing NEMO) fail because of: "psyclone/parse/module_manager.py", line 219, in get_modules_in_file source_code = file_in.read() ^^^^^^^^^^^^^^ File "", line 322, in decode UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd0 in...