Adding support for equi-join on struct

[hyperbolic2346]

Adds support for equijoin on structs.

This PR is leveraging the struct PR and the rewrite for join API. It enables equijoin on structs by flattening the struct for the hash calculation.

closes #7543

[hyperbolic2346]

This is now ready for review.

[harrism]

This PR is leveraging struct PR and rewrite for join API.

Not sure what you mean by this. Please clarify with readers of the changelog in mind, since this will get pasted into it.

[codecov]

Codecov Report

Merging #7720 (2b0d7ee) into branch-0.19 (7871e7a) will increase coverage by 0.78%.
The diff coverage is n/a.

❗ Current head 2b0d7ee differs from pull request most recent head b946217. Consider uploading reports for the commit b946217 to get more accurate results

@@               Coverage Diff               @@
##           branch-0.19    #7720      +/-   ##
===============================================
+ Coverage        81.86%   82.65%   +0.78%     
===============================================
  Files              101      103       +2     
  Lines            16884    17524     +640     
===============================================
+ Hits             13822    14484     +662     
+ Misses            3062     3040      -22     

Impacted Files	Coverage Δ
python/cudf/cudf/utils/dtypes.py	83.27% <0.00%> (-6.24%)	⬇️
python/cudf/cudf/utils/gpu_utils.py	53.65% <0.00%> (-4.88%)	⬇️
python/cudf/cudf/core/column/lists.py	87.32% <0.00%> (-4.08%)	⬇️
python/dask_cudf/dask_cudf/backends.py	87.50% <0.00%> (-2.13%)	⬇️
python/cudf/cudf/core/abc.py	87.23% <0.00%> (-1.14%)	⬇️
python/cudf/cudf/core/column/decimal.py	93.84% <0.00%> (-1.03%)	⬇️
python/cudf/cudf/core/column/column.py	87.53% <0.00%> (-0.23%)	⬇️
python/cudf/cudf/utils/utils.py	85.36% <0.00%> (-0.07%)	⬇️
python/cudf/cudf/io/feather.py	100.00% <0.00%> (ø)
python/cudf/cudf/utils/ioutils.py	78.71% <0.00%> (ø)
... and 49 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 684bb14...b946217. Read the comment docs.

[hyperbolic2346]

Not sure what you mean by this. Please clarify with readers of the changelog in mind, since this will get pasted into it.

Am I correct in thinking that only the first line is used in the changelog? I have updated the comment to have a one-liner for what the change is on a high level and then detail for reviewers below that.

[nvdbaranec]

Surprisingly simple. Does flatten_nested_columns throw an exception if it encounters a list column?

[hyperbolic2346]

Surprisingly simple. Does flatten_nested_columns throw an exception if it encounters a list column?

It marches the table columns looking for structs. Doesn’t complain about lists.

if (child.type().id() == type_id::STRUCT)

[nvdbaranec]

Surprisingly simple. Does flatten_nested_columns throw an exception if it encounters a list column?

It marches the table columns looking for structs. Doesn’t complain about lists.

if (child.type().id() == type_id::STRUCT)

So if you had this:

struct<int, struct<list<struct<...>> , list, struct<int>>>>

it would flatten to

int, list, list, int ? That first list quietly contains a struct within it. Does that mess things up? Or is it the case that the mere presence of list columns in the flattened output will cause the join to throw later on?

[karthikeyann]

@nvdbaranec I am working on templatized code for flatten_nested_columns. (still doesn't have list support, but throws error). Targeted for 0.20 release.

[hyperbolic2346]

@nvdbaranec I just coded this up as a quick test and we get "C++ exception with description "cuDF failure at: ../../../../include/cudf/table/row_operators.cuh:363: Attempted to compare elements of uncomparable types." thrown. This is the same error as passing a list by itself to join.

[harrism]

Not sure what you mean by this. Please clarify with readers of the changelog in mind, since this will get pasted into it.

Am I correct in thinking that only the first line is used in the changelog? I have updated the comment to have a one-liner for what the change is on a high level and then detail for reviewers below that.

Nope, not correct. :)

[devavret]

This is such a neat little PR.

[hyperbolic2346]

@gpucibot merge

[hyperbolic2346]

@gpucibot merge