combinatorics coverage

[Ismael-VC]

Test next for zero length permutation.

[nalimilan]

Thanks. Wouldn't it be more logical to use collect(permutations([])), though?

[tkelman]

@nalimilan could you elaborate? how about testing both?

[nalimilan]

Why not testing both, but next is tested by collect AFAIK, and the current code depends on the implementation detail of what the iterator state is. We don't really care if it changes, as long as iteration (and thus collect) still works.

[Ismael-VC]

@nalimilan I will change it according to your suggestions.

Today I will try to set up my fork of Julia to work with Travis/CoverAlls. Since right now I don't get "immediate" feedback of whether a test actually covered some code-path or if it didn't. It seems that the Julia build-bots used for coverage are only run once or twice a day.

That will make this quest easier!

[Ismael-VC]

Another option is to learn to use the "manual"command line coverage tools, but for the moment I'd rather use CoverAlls, so I can keep focusing in learning git and Julia instead of yet another command line tool.

[Ismael-VC]

@nalimilan Is this correct?

[nalimilan]

Looks good to me. AppVeyor failure is unrelated, I don't think it's worth restarting it.

@tkelman Is it fine with you?

[nalimilan]

Oh, and could you squash the commits?

[Ismael-VC]

@nalimilan squashing done!

[tkelman]

cancelled appveyor, it will probably time out due to #11818 - we can restart the build later if that gets fixed, though there's likely to be a worse-than-usual backlog for the next few days

[nalimilan]

I would just merge it, I don't see why it would fail on AppVeyor and not on Travis.