-
Notifications
You must be signed in to change notification settings - Fork 16
Open
Description
Looking through the biggest culprits of long testing times I found test_model_compile
which on my m4 macbook pro takes just over one hour to run each (topquadrant and pyshacl) test. This test is fairly basic, it compiles a model and compares it to a pre-compiled model.
What is taking so long?
The actual compilation step of the process only takes about 30 seconds. However, the action of comparing the compiled and pre-compiled model takes an hour. This comes down to the triple canonicalization step needed to compare the models.
What can we do to make this test shorter?
- Could we use a less complicated example case?
- Could we use a different method for comparison?
- Could we use a heuristic approach to comparison?
- Could we remove some triples before comparing, which we know to be the same?
Metadata
Metadata
Assignees
Labels
No labels