Moving some slow tests to integration#1442
Conversation
andrewlee94
requested review from
MAZamarripa,
adowling2,
bpaul4,
jsiirola,
ksbeattie,
lbianchi-lbl and
rundxdi
as code owners
andrewlee94
added
Priority:Normal
| def test_general_hx1d_initializer(self, model): | ||
| initializer = HX1DInitializer() | ||
| initializer.initialize(model.fs.unit) |
There was a problem hiding this comment.
I wonder how this initialization time is going to be affected when we move away from block triangularization in #1436 .
There was a problem hiding this comment.
That is a good point, but I suspect it won't have a huge effect in this case. I believe there is another equivalent test for a different configuration that runs fast enough, so I suspect this one is caused more by stiffness in the model than the property initialization. Still, would be worth waiting for #1436 to check.
There was a problem hiding this comment.
Update: after #1436 merged, run time for this test is still ~16 seconds, so I'll leave it in integration for now.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1442 +/- ##
==========================================
- Coverage 77.74% 76.35% -1.39%
==========================================
Files 394 394
Lines 64953 64953
Branches 14404 14404
==========================================
- Hits 50497 49598 -899
- Misses 11875 12797 +922
+ Partials 2581 2558 -23 ☔ View full report in Codecov by Sentry. |
4 participants
Fixes None
Summary/Motivation:
Run times for our non-integration tests have been creeping up again, so this PR moves a lot of slow tests (>10s) to integration. This will likely impact our apparent test coverage, as this is calculated based on the
unitandcomponentflags only, but cuts the run time for these by nearly 30% on my local machine.Legal Acknowledgement
By contributing to this software project, I agree to the following terms and conditions for my contribution: