More error checking in from_dlpack

[wence-]

The implementation copying a full column at a time relies on
unit-stride, column-major data in the incoming dlpack tensor. Check
these preconditions and report error messages, rather than either
silently producing bad data (unit-stride, row-major 2D tensors) or
invalid memory accesses (non-unit-stride in the fastest-varying
dimension). Closes #10754 by providing an explicit error message for
this unsupported case.

[wence-]

I would like to add some tests for this, but need a little guidance on where the best place is/how to do so. I guess since cupy is a hard dependency I can just test at the python level?

[PointKernel]

@wence- I'm not familiar with this part of the code but looks like we can custom a tensor (https://github.com/rapidsai/cudf/blob/branch-22.06/cpp/tests/interop/dlpack_test.cpp#L345-L351) and verify if the function can throw the expected message via CUDF_EXPECT_THROW_MESSAGE (a simple use case here).

[wence-]

@wence- I'm not familiar with this part of the code but looks like we can custom a tensor (https://github.com/rapidsai/cudf/blob/branch-22.06/cpp/tests/interop/dlpack_test.cpp#L345-L351) and verify if the function can throw the expected message via CUDF_EXPECT_THROW_MESSAGE (a simple use case here).

Thanks, although I am loathe to introduce testing of specific error messages, in light of #10632 and the in-progress #10637.

[PointKernel]

@wence- I'm not familiar with this part of the code but looks like we can custom a tensor (https://github.com/rapidsai/cudf/blob/branch-22.06/cpp/tests/interop/dlpack_test.cpp#L345-L351) and verify if the function can throw the expected message via CUDF_EXPECT_THROW_MESSAGE (a simple use case here).

Thanks, although I am loathe to introduce testing of specific error messages, in light of #10632 and the in-progress #10637.

EXPECT_THROW may be what you need then.

[codecov]

Codecov Report

Merging #10850 (4b9743f) into branch-22.06 (9def28c) will increase coverage by 0.09%.
The diff coverage is n/a.

@@               Coverage Diff                @@
##           branch-22.06   #10850      +/-   ##
================================================
+ Coverage         86.23%   86.32%   +0.09%     
================================================
  Files               143      144       +1     
  Lines             22546    22656     +110     
================================================
+ Hits              19442    19558     +116     
+ Misses             3104     3098       -6     

Impacted Files	Coverage Δ
python/dask_cudf/dask_cudf/backends.py	85.51% <0.00%> (-0.94%)	⬇️
python/cudf/cudf/core/scalar.py	89.01% <0.00%> (-0.31%)	⬇️
python/dask_cudf/dask_cudf/io/tests/test_s3.py	96.05% <0.00%> (-0.15%)	⬇️
python/cudf/cudf/io/__init__.py	100.00% <0.00%> (ø)
python/cudf/cudf/core/udf/typing.py	97.54% <0.00%> (ø)
python/dask_cudf/dask_cudf/tests/test_groupby.py	100.00% <0.00%> (ø)
python/cudf/cudf/core/missing.py	100.00% <0.00%> (ø)
python/cudf/cudf/core/column/column.py	89.77% <0.00%> (+0.01%)	⬆️
python/cudf/cudf/core/column/numerical_base.py	98.91% <0.00%> (+0.01%)	⬆️
python/cudf/cudf/core/_internals/where.py	94.73% <0.00%> (+0.03%)	⬆️
... and 13 more

---

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 9def28c...4b9743f. Read the comment docs.

[wence-]

EXPECT_THROW may be what you need then.

Thanks, done.

[davidwendt]

There are a couple of instances (lines 169 and 174) near these changes where the CUDF_EXPECTS text reads
greater than or equal-to
Could you remove the hyphen?

[wence-]

There are a couple of instances (lines 169 and 174) near these changes where the CUDF_EXPECTS text reads greater than or equal-to Could you remove the hyphen?

Done in 4b9743f

[PointKernel]

@gpucibot merge