tests are failing in ARAX pytest suite since the KP TRAPI version was set to 1.6.0 in the master branch

[saramsey]

Hi @edeutsch, just an FYI, commit 83ed888 seems to have broken some ARAX pytests, in the master branch code.

I'm running the pytest suite like this, on my MBP:

source venv/bin/activate
cd RTX/code/ARAX/ARAXQuery/Expand
python -m kp_info_cacher
cd ../..
pytest -v --cache-clear test/

With the code in the master branch in the previous commit (b4f9cc7), running the pytest suite on my MBP, of the 154 default tests in the master branch, all 154 were passing:

Now, when I check out commit 83ed888 from the mast er branch, only 93 are passing, and 61 are failing:

[saramsey]

I am going to work on fixing the unit tests, in branch issue-2712, off of the current HEAD of the master branch (which I have also verified is failing 61 pytests).

[saramsey]

Tentatively marking this issue high priority since I feel that we need the pytests to be back in a passing state soon, in order to facilitate future development (in order to ensure no regression of standards-compliance or ARAX capabilities, in future features or bugfixes beyond 83ed888). But if others feel this is not really a high priority issue, please let me know.

[edeutsch]

ah apologies @saramsey, I did not think to run the test suite for this simple update, but in retrospect it seems obvious that it would probably cause mayhem.

I had opined that it would be nice to keep the Legacy ARAX and the Refactor ARAX codebases unified since the only change was one character. But it seems that in practice perhaps the codebases cannot be easily unified if so many of the tests depend on changing data?

Or is it possible to adjust the tests so that they work under both scenarios?

Or should we just have somewhat different tests for Legacy vs. Refactor?

Maybe we can keep the main codebase the same but different tests?

Sorry for the trouble.

[saramsey]

I have the unit tests for ARAX_expand working again (see commit 27b5e1a for all of the tweaks required for this).

[saramsey]

I am down to only three failing unit tests in test_ARAX_resultify.py....

[saramsey]

With commit 68b4e5f, all 29 unit tests in test_ARAX_resultify.py are now passing.

[saramsey]

These tests in test_ARAX_overlay.py are failing:

=========================================================== short test summary info ============================================================
FAILED test/test_ARAX_overlay.py::test_jaccard - AssertionError: assert 'ERROR' == 'OK'
FAILED test/test_ARAX_overlay.py::test_add_node_pmids - AssertionError: assert 'ERROR' == 'OK'
FAILED test/test_ARAX_overlay.py::test_compute_ngd_virtual - AssertionError: assert 'ERROR' == 'OK'
FAILED test/test_ARAX_overlay.py::test_compute_ngd_attribute - AssertionError: assert 'ERROR' == 'OK'
FAILED test/test_ARAX_overlay.py::test_FET_ex1 - AssertionError: assert 'ERROR' == 'OK'
=================================================== 5 failed, 20 skipped, 1 warning in 1.08s ===================================================

[saramsey]

I am fixing the tests in the issue-2712 branch as one-way-trip, meaning, I am fixing them to work only with infores:retriever.

It would be difficult to make them able to work with either 1.5.0 (meaning mostly using infores:rtx-kg2 for many of the KP-restricted unit tests) or with 1.6.0 (meaning exclusively using infores:retriever, and never using infores:rtx-kg2). For legacy ARAX, perhaps some kind soul can archive the old unit tests in some other directory, other than RTX/code/ARAX/test.

[saramsey]

With commit 0cd0b55, all unit tests in test_ARAX_overlay.py are now passing.