Get IntervalArithmetic.jl ready for Julia 1.0

[dpsanders]

(Incorporates the previous PRs for 0.7.)

• setrounding will be handled by https://github.com/JuliaIntervals/SetRounding.jl once it is registered (in process).

• New iteration protocol.

• Add global in loops in testsets

• Reenable complex tests. Complex powers do not work Powers of complex intervals are broken on Julia 0.7 #198

[dpsanders]

Sorry, should have rebased instead of merging.

[dpsanders]

I suggest that we don't squash when we merge this.

[coveralls]

Coverage increased (+4.4%) to 97.104% when pulling c0e96cf on dps/0.7 into a9d3ef0 on master.

[codecov-io]

Codecov Report

Merging #194 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #194   +/-   ##
=======================================
  Coverage   92.67%   92.67%           
=======================================
  Files          24       24           
  Lines        1120     1120           
=======================================
  Hits         1038     1038           
  Misses         82       82

---

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 a9d3ef0...c0e96cf. Read the comment docs.

[dpsanders]

Tests are passing! ;) Still a bit of work to do though.

[lbenet]

Note that deprecation warnings in CRlibm were solved; it is a question of tagging a new version.

[lbenet]

Nice work!

[lbenet]

I just checked that, except for a single test, all tests in test/interval_tests/complex.jl pass. The problematic one is @test a^2 == -1. I suggest to have the tests for complex numbers, and either comment the problematic case or use @test_broken for it.

[lbenet]

@mohamed82008 Have you check that tests pass in windows for this branch (dps/0.7)?
If I recall some of the discussion, in spite of appveyor opinion, tests were failing in windows.

[mohamed82008]

Not yet. I will as soon as I'm infront of a computer.

[dpsanders]

@lbenet Great, let's do that then. Would you like to do it or shall I?

[mohamed82008]

Tests pass on my machine. I am only getting the setrounding warning.

[dpsanders]

The latest commit needs the next tagged version of CRlibm (in progress) and of SetRounding (to come).

[dpsanders]

All tests are passing on 0.7, 1.0 and nightly, on all platforms!
This is ready to merge. cc @lbenet

[lbenet]

Thanks a lot for the big effort!

I think we should first tag a last version which is compatible with Julia 0.6 of ValidatedNumerics.jl and the rest of the packages with the proper version, and then merge and tag this PR. See JuliaIntervals/ValidatedNumerics.jl#285, whose tests should be working now.

[dpsanders]

There are no changes on master since v0.14.0, so we can treat v0.14.0 as the last version compatible with Julia 0.6.

[dpsanders]

Upper bounds for the other packages can be added retroactively directly in METADATA, without tagging new versions of them.

[lbenet]

Shall we merge this and tag a new version (Julia 0.7 and 1.0 compatible only)?

[dpsanders]

Merging...