Closed anilyil closed 2 years ago
Merging #114 (f6a8fb4) into master (1344987) will decrease coverage by
1.09%
. The diff coverage is100.00%
.:exclamation: Current head f6a8fb4 differs from pull request most recent head 5c2b7de. Consider uploading reports for the commit 5c2b7de to get more accurate results
@@ Coverage Diff @@
## master #114 +/- ##
==========================================
- Coverage 60.43% 59.34% -1.10%
==========================================
Files 40 40
Lines 9671 9669 -2
==========================================
- Hits 5845 5738 -107
- Misses 3826 3931 +105
Impacted Files | Coverage Δ | |
---|---|---|
pygeo/geo_utils/remove_duplicates.py | 76.63% <100.00%> (-0.22%) |
:arrow_down: |
pygeo/constraints/areaConstraint.py | 51.63% <0.00%> (-24.84%) |
:arrow_down: |
pygeo/constraints/DVCon.py | 63.97% <0.00%> (-3.24%) |
:arrow_down: |
pygeo/parameterization/DVGeo.py | 65.67% <0.00%> (-0.36%) |
:arrow_down: |
pygeo/geo_utils/file_io.py | 65.78% <0.00%> (-0.23%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 1344987...5c2b7de. Read the comment docs.
Purpose
This PR addresses #89. The underlying cause was an argsort flipping the order of some entries due to the differences on the order of machine precision. By rounding the array to 8 significant figures before the sorting algorithm, we prevent this behavior.
This might cause some existing runscripts to get a different ordering but as we saw with the test case that started failing, this means that these runscripts were susceptible to this issue anyways. If it shows up, this change will show up as a shuffled ordering of the FFD control points in shape variables.
Type of change
Testing
We'll see what azure says
Checklist
flake8
andblack
to make sure the code adheres to PEP-8 and is consistently formatted