change real to float in cor of a single collection

[bkamins]

As in cor we get square root I think it is safe to assume that the result should be floating point.

An example of current surprising behavior:

julia> cor([im])
true

[nalimilan]

Thanks!

[nalimilan]

Though could you add a test to check the new behavior?

[bkamins]

A good point - I have added the tests. They highlight the cases that I think are critical. A check that in all cases we produce what we think should be produced is welcome (I think they are not always intuitive, but I made the PR to be non-breaking in these corner cases except changing the type of 1 produced).

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@7a0c0d1). Click here to learn what that means.
The diff coverage is n/a.

@@           Coverage Diff            @@
##             master     #61   +/-   ##
========================================
  Coverage          ?   0.00%           
========================================
  Files             ?       1           
  Lines             ?     381           
  Branches          ?       0           
========================================
  Hits              ?       0           
  Misses            ?     381           
  Partials          ?       0           

---

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 7a0c0d1...c6bfe2f. Read the comment docs.

[nalimilan]

I just noticed corm and corzm have similar definitions. I guess these should be changed too, right?

[nalimilan]

@stevengj You introduced this definition at JuliaLang/julia#13348, replacing float(one(T) * one(T)) with real(one(T)). Yet cor([1im, 2im], [1im, 2im]) returns 1.0 + 0.0im, not 1.0. I guess nobody really tested nor care about this use case. What do you think we should do? We could also use float(real(one(T)) to minimize breakage; what matters is that cor([1, 2]) returns 1.0, not 1.

[bkamins]

I guess these should be changed too, right?

What do you mean exactly? I have already changed corm and corzm in this PR (and added tests for this).

[nalimilan]

What do you mean exactly? I have already changed corm and corzm in this PR (and added tests for this).

Woops, forget it, I was just confused.

[nalimilan]

@stevengj Any chance you could find the time to look at this?