[REVIEW] Port unary cast to libcudf++

[trxcllnt]

Port unaryops cast for issue #2950, related to @codereport's PR #3214.

[codecov]

Codecov Report

Merging #3397 into branch-0.11 will not change coverage.
The diff coverage is n/a.

@@             Coverage Diff              @@
##           branch-0.11    #3397   +/-   ##
============================================
  Coverage        87.36%   87.36%           
============================================
  Files               49       49           
  Lines             9295     9295           
============================================
  Hits              8121     8121           
  Misses            1174     1174

---

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 7f5e054...189e68a. Read the comment docs.

[harrism]

Thanks for contributing this, @trxcllnt

[codereport]

Looks good! (pending resolved CI failures).

[trxcllnt]

rerun tests

[trxcllnt]

@jrhemstad I cleaned up the implementations and changed the signatures to return unique_ptr<column>. Now that #3232 is merged, I was thinking of updating this PR to call to_timestamps and from_timestamps when casting TS to/from strings. Does that sound like the right approach?

[jrhemstad]

I was thinking of updating this PR to call to_timestamps and from_timestamps when casting TS to/from strings. Does that sound like the right approach?

If that's the expected behavior of casting to/from strings, then sure.

[trxcllnt]

It would be convenient to have a single cudf::cast() method to convert columns to any other type, but after looking through the methods in cudf/strings/convert, I've decided against it.

Each string method takes different parameters based on the to/from type, and it isn't possible to define cast() overloads for each convert func parameter list. Better to guide people to those functions explicitly when they want to go to/from strings.

[harrism]

What's the status of this PR?

[trxcllnt]

@harrism to answer your question, now that I've updated the cast docstring I think this is ready to merge.