Fixed.

The internal solution is based on regular expressions and the assumption that the input array only contains empty strings and words (as in regex \w or [a-zA-Z_0-9]), so it generally fails on strings like e.g. start-up. When combined with the random word generation that mistakenly puts some { character(s) in the last word on occassion the internal solution fails to use " and " as the last delimiter and falls back to ", ".

That means: If the last word contains an opening curly brace don't use " and " as the last delimiter but instead use ", ".

Still getting this error on 17-11-2017
So confused :(

Still getting this error on 10/11/2017

Still not fixed...!!

But it is still not fixed!

Okay, I found the issue. It's actually with the way the test cases generate random words, although, I am unable to edit it.

For reference. In the generateRandomWord() function in the test cases, rnd.nextInt((123 - 97) + 1) + 97)) should be: rnd.nextInt((122 - 97) + 1) + 97)).

Sorry guys, that was my fault.

no, the solution doesn't work:

words = [mq{bmlwrk, , ja, vqzbrfe, kpdqp{c, w, zgizqv, sz{u, tt, chxqq, bvgo, o, dkzzb, l, jvwt, ovfg{{]
expected:<...o, o, dkzzb, l, jvwt[,] ovfg{{> but was:<...o, o, dkzzb, l, jvwt[ and] ovfg{{>

• I do not mutate the array
• actual and expected are not swapped so the "but was" WITH the "and" is effectively my answer...
• so the internal solution IS wrong.

I just tested the solution and it works.

I've had issues with codewars before where the expected and actual results were the wrong way round.

Please fix the translation - see in discourse.

Same problem in java

expected:

<...of, wuyozz, wcl{qouh[,] gdu{rbwg>

but was:

<...of, wuyozz, wcl{qouh[ and] gdu{rbwg>

so adding and in the end is wrong????????? Im so confused

?????????????????

How did this get pass the test?