Simplified traverse method for vectors by using foldLeft

[amitayh]

By doing this, there's no need for the Eval wrapper and avoiding the
foldRight which could potentially be more efficient

[codecov-io]

Codecov Report

Merging #2091 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2091      +/-   ##
==========================================
- Coverage   94.63%   94.63%   -0.01%     
==========================================
  Files         318      318              
  Lines        5391     5390       -1     
  Branches      209      210       +1     
==========================================
- Hits         5102     5101       -1     
  Misses        289      289

Impacted Files	Coverage Δ
core/src/main/scala/cats/instances/vector.scala	72.09% <100%> (-0.64%)	⬇️

---

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 3e1a053...47cb3fd. Read the comment docs.

[ceedubs]

Thanks for bringing this up, @amitayh.

It could potentially be more efficient, but foldLeft is also always going to do a full traversal of the vector. For example consider the code (1 to 100000).toVector.traverse[Option, Int]{i => println(i); None}. With the current implementation, it will print 1 and then halt. With your change it will print the full range.

I actually thought that we had some tests that were checking that traverse was short-circuiting, but it appears that we don't at least for Vector. While there are definitely trade-offs between the two approaches, I'm inclined to think that the current implementation is probably a better default. I think that we should probably add some tests to ensure the traverse is short-circuiting the structure when possible.

[ceedubs]

It looks like we have these tests for List, Stream, and a couple of other types. I'll submit a PR to add them for Vector.

[johnynek]

Can we add benchmarks when we are making such trade offs?

[amitayh]

Is there some tool you're using for benchmarking? I can try to run a few tests...

[ceedubs]

@amitayh there are some benchmarks in https://github.com/typelevel/cats/tree/master/bench/src/main/scala/cats/bench that would probably be a good starting point.