To test to effects of PR #70, I've added a codegen test where P is None. This would have failed on Windows before PR #70 but should pass now. Specifying P as None and later updating to a non-None value using the update_P method does not seem to be supported, so I'm only testing some other cases here - solve, update_A etc.
While I was at it, I noticed that the codegen tests were relying on the fact that test_solve will be executed before any other tests (test_update_*). This is not guaranteed but is probably only happening because tests are picked up in alphabetic order, so I've moved the codegen logic to the class method, and am re-setting the model in setUp.
To test to effects of PR #70, I've added a codegen test where
PisNone. This would have failed on Windows before PR #70 but should pass now. SpecifyingPasNoneand later updating to a non-None value using theupdate_Pmethod does not seem to be supported, so I'm only testing some other cases here -solve,update_Aetc.While I was at it, I noticed that the codegen tests were relying on the fact that
test_solvewill be executed before any other tests (test_update_*). This is not guaranteed but is probably only happening because tests are picked up in alphabetic order, so I've moved the codegen logic to the class method, and am re-setting the model insetUp.Also enabled codegen checks on the CI.