Open trev-f opened 2 months ago
One potential solution is to create process tests only during development that don't get integrated into the full testing suite. This would have the benefit of allowing for testing during development without cluttering up new pipelines with failing tests.
Subworkflows or even workflows could then be treated as the units for unit testing. This would mean there are far less changes to make to the test suite when a new pipeline is forked off of ngs. The other benefit is faster runs of the full test suite.
Unit testing each process is great for making sure we get expected outputs during development. However, all of these unit tests really end up hampering development time when a new pipeline is forked off of this one and has a different input structure.
Is there some good way to keep these unit tests from becoming so ossified?