- Mock requirements that are not needed using 'autodoc_mock_imports'
- Add necessary requirements to requirements.txt
- These changes should mean the docs can be built without first
installing gprMax
- Some of the ReFrame tests fail on ARCHER2. This is due to differences
in floating point number errors in the MPI and non-MPI runs.
- FineGeometryViews fail as the tests do not support checking VTKHDF
unstructured grids with parallel partitions.
- Better variable names to hopefully make it clearer how the algorithm
works
- Reduce the number of calls to `rng.standard_normal()` by combining
calls that were at the start and ends of the for loops
- Remove unnecessary calls to `rng.standard_normal()` at the end of the
final iteration of the for loops
- Reduce the size of the regression files for testing geometry objects
- Don't save geometry objects when testing geometry commands. This means
the regression check is done using the receiver output rather than
reading the geometry directly. Some tests still output the geometry
for reference (useful if tests fail to check the geometry that was
built).
- When some ranks finish quicker than others, if the ranks one of the
ranks still running encounters an error it will attempt to call
MPI.Abort. As some ranks will have already exited, this was causing
deadlock without the error actually appearing.
- 'output_file' is not stored in the run report by reframe. Therefore,
trying to access 'test_dependency.output_file' where 'test_dependency'
is a restored test case was returning None and causing an error.
- Instead, created a new function to build an output file path that
can be reused by the ReceiverMixin class.
- Adding cell and point data is very dependent on the type of the VTKHDF
file. Therefore these should be private in the base class so an
instance of the VtkHdfFile class can only add field data.
