Note that some extra 0's may naturally occur at the beginning and the end of a message, make sure to ignore them.

Those 0s are not a reference. After disregarding them, you end up with "111" which is "E".

Fixed for Java.

That looks like an error in the tests, mention the language you're talking about, it doesn't happen in javascript.

No, there is an if inside the second loop to avoid that.

Use spoiler flag next time, please. Your post was visible in the homepage.

Please use spoiler flag next time.

Please use spoiler flag next time, your post was visible in the homepage.

Approved.

Haskell fork that uses approximate equality

I think you meant a comment, it can be done if it has no replies, but don't worry, just make sure you use spoiler flag when needed.

Please use spoiler flag next time, your post was visible in the homepage.

Please don't post solutions in kata's discourse without marking your post as having spoiler content, even if they don't work, read this: https://docs.codewars.com/training/troubleshooting#post-discourse

About your problem, your code is too slow and does too many operations, that's why it times out. Nested loops, yes filter is a loop in disguise, won't work here.

Sorry, but that seems to be a problem on your part, not in the instructions. I've changed the for a, but I still don't see how you could read this:

i thought that only 15 should be added to the sum excluding all commom multiple of 3 and 5.

 If a number is a multiple of both 3 and 5, only count it once.

         ^---------------------------------------------^       

15 is a multiple of both 3 and 5, but you add it only once, the same for 30, 45 and so on.

Given the issue below, saying that this kata is a duplicate, I wouldn't waste my time with it, I don't think it will ever be published. Unless the second point you mentioned would make it unique.