Closed ianthomas23 closed 1 year ago
Merging #1205 (8b7a2a4) into main (80d1ccf) will decrease coverage by
0.08%
. The diff coverage is75.00%
.
@@ Coverage Diff @@
## main #1205 +/- ##
==========================================
- Coverage 84.70% 84.63% -0.08%
==========================================
Files 35 35
Lines 8357 8354 -3
==========================================
- Hits 7079 7070 -9
- Misses 1278 1284 +6
Impacted Files | Coverage Δ | |
---|---|---|
datashader/core.py | 88.25% <ø> (-0.13%) |
:arrow_down: |
datashader/glyphs/glyph.py | 81.72% <50.00%> (ø) |
|
datashader/resampling.py | 82.09% <100.00%> (-1.03%) |
:arrow_down: |
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
There is a warning
<snip>/site-packages/numba/np/ufunc/dufunc.py:84: NumbaDeprecationWarning: The 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.
dispatcher = jit(_target='npyufunc',
which can be ignored as it was fixed in https://github.com/numba/numba/pull/8909 and the fix will be included in the numba 0.57
release.
I am happy with this now, all tests are passing with new numba 0.57
, numpy 1.24
and python 3.11
as well as previous versions of all 3. The CI failures are on codecov which is the usual inability to determine coverage of numba
code.
The only change that could be removed is downloading packages from the numba
conda channel; I will remove this in a separate PR when numba 0.57
is available on conda-forge
.
This is a WIP to support numba 0.57. We can no longer pass masked arrays to numba, but that is OK as we handle the masks separately anyway.
There is a separate problem that occurs on macOS but not linux, that I am still investigating.