Transpose

[RobinCsl]

There is a commit comment relating to files which should have been in their own branch, but I worked on master first, and then realised that to make a second PR, I should have used a separate branch... Hopefully that's not a problem!

[matthewmorgan]

@RobinCsl no worries at all about working on master-- since you're on your own fork, it's not a problem. I'll review.

[matthewmorgan]

@RobinCsl if you could please change the language I mentioned in the README, and have a look at my comments on the code suggestions. Code suggestions are entirely optional/food for thought. Let me know how you want to proceed.

[matthewmorgan]

Same language issue I mentioned on your other PR. All tests have been skipped except the first, so might as well say so.

[matthewmorgan]

   [...line].forEach((value, key) => {

The ... is the "spread" operator. It's awesome.

[matthewmorgan]

This all looks good, but it did make me wonder if it's possible to do this entirely functionally-- IE, without having an initial result array that gets mutated. Not critical, but just thinking out loud...

[RobinCsl]

I updated the README and used the spread operator (it's really cool indeed!). I tried briefly to think about rewriting the example in a purely functional way, but to no avail at the moment.

[matthewmorgan]

Looks fine to me! Mind squashing the commits?

[RobinCsl]

It seems I somehow managed on this one. I'll try to replicate the steps. (But there might be other conflicts since each exercise implementation modifies config.json)

[rchavarria]

@matthewmorgan approved changes, commits squased, no conflicts,... merging.

[shybyte]

@RobinCsl The expected test result in the last test ("test many lines") is different from the expected test result in https://github.com/exercism/problem-specifications/blob/master/exercises/transpose/canonical-data.json

So either the canonical test-data is wrong or this PR is wrong.

[matthewmorgan]

@shybyte good catch! @RobinCsl I need to remove this exercise for the time being. Please have a look and let me know if I can help resolving the issue.

[RobinCsl]

I had a look for the last hour. I had copy-pasted the code from the old repository and converted it to ES6, but you're right @shybyte, there is an issue with the algorithm. I will work more on that.

[shybyte]

@RobinCsl Maybe my code for solving the canonical testcases is helpful:

function trimTrailingUndefined(array) {
  const trailingUndefinedCount = [...array].reverse().findIndex(x => x !== undefined);
  return array.slice(0, array.length - trailingUndefinedCount);
}

export default function transpose(input) {
  const maxCol = Math.max(0, ...(input.map(row => row.length)));
  return [...Array(maxCol).keys()].map(col =>
    trimTrailingUndefined(input.map((_v, row) => input[row][col]))
      .map(charOrUndefined => charOrUndefined || ' ')
      .join('')
  );
}