[REVIEW] Enable fixed_point binary operations

[codereport]

This addresses are part of #3556. It is a follow up to #5861 where most of the work was done. However, this PR fixes an issue with binary_operator::DIV.

Starting point is to comment in the commented out unit tests in binop-integration-tests.cpp and go from there.

Update:

After #6544 got merged, this job was simplified to the following:

• Enable bool is_supported_fixed_point_binop(binary_operator op)
• Uncomment out tests
• Modify FixedPointBinaryOpDiv test
• Add a couple more binary_operator::DIV tests

[GPUtester]

Please update the changelog in order to start CI tests.

View the gpuCI docs here.

[codecov]

Codecov Report

Merging #6528 into branch-0.17 will not change coverage.
The diff coverage is n/a.

@@             Coverage Diff              @@
##           branch-0.17    #6528   +/-   ##
============================================
  Coverage        82.72%   82.72%           
============================================
  Files               90       90           
  Lines            14892    14892           
============================================
  Hits             12320    12320           
  Misses            2572     2572           

---

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 17cd94f...cfc6d30. Read the comment docs.

[harrism]

Please target branch-0.17.

[sperlingxx]

For current implementation, it seems that fixed_point types are not able to perform binary opeartion with other numeric types. Will we enable it in this ongoing PR?

[codereport]

For current implementation, it seems that fixed_point types are not able to perform binary opeartion with other numeric types. Will we enable it in this ongoing PR?

No, that is not a part of this PR.

[harrism]

@sperlingxx Once (unary) casts are implemented the approach would be to cast to fixed point and then do the binary operation.

[sperlingxx]

@sperlingxx Once (unary) casts are implemented the approach would be to cast to fixed point and then do the binary operation.

I got it. Thanks, Mark!

[devavret]

This is from a previous PR but during the creation of temporary result in fixed_point_binary_operation, cudf::binary_operation is being used to create a temporary column of the same scale. This does not take a stream. Try using the detail version.

[codereport]

This job is currently waiting on #6544 to be merged.