Make async.transform actually support 2 arguments

[webbiesdk]

The documentation suggests that both the accumulator and callback arguments to transform is optional.
(https://caolan.github.io/async/docs.html#transform)

However, if both are left out (and transform therefore is called with only 2 arguments), transform crashes with a iteratee is not a function exception.

By changing the equality check to a less than check, the transform function support the use-case where only 2 arguments are passed.

[webbiesdk]

I have no idea why the two builds are failing.

The coverage test claims that some internal filter.js has less coverage. Even though my change doesn't affect that.

And somehow a test-case failed, after I added a test-case in another file.

[webbiesdk]

I just triggered a rebuild by closing and re-opening the issue, and now the tests are passing.

You guys have an unstable test!

[megawac]

Yeah, its a reoccurring issue (#1322) sorry about that.

As for the PR, sure transform could support taking a source to iterate and an iteratee method without a final callback but what would be the point?

[megawac]

Also the test failing looks to be coveralls complaining that the coverage of the repo decreased for some reason

[webbiesdk]

As for the PR, sure transform could support taking a source to iterate and an iteratee method without a final callback but what would be the point?

The point would be that your current documentation states that callback is optional.
The current implementation actually supports an optional callback, but only if the number of arguments is exactly 3.

[webbiesdk]

Also the test failing looks to be coveralls complaining that the coverage of the repo decreased for some reason

The test that failed until i triggered the rebuild was: retry with interval when some attempts fail and error test returns false at some invokation (but it is already marked as flaky in the test-source).

I also tried to add a new test to cover a bit of the uncovered code, and now the coverage result makes no sense, since it states that every file with changed coverage has the same or increased coverage (but somehow the overall coverage is still down).

I also tried to run npm run coverage with and without my change, and it reports the exact same coverage in both cases.

The decreased coverage is caused by a change in the environment.
To confirm that i looked at the last coverage report that gave high coverage. Just try to look at these two:

https://coveralls.io/builds/10391324/source?filename=lib%2Finternal%2FsetImmediate.js (the last time coverage was run on master)
https://coveralls.io/builds/10579695/source?filename=lib%2Finternal%2FsetImmediate.js (coverage test after my change)

[megawac]

Thanks, yeah the retry test failing is a known issue. Thanks for pointing out the docs were inaccurate.

I'll look into this coverall issue issue later, seems odd its complaining when you've added tests.

Thanks for the pull request!