Michel Hermier

Yes, and a little bit more than that, if there was *legitimate* invalid chars in the middle of the sequence they would have bin previously silently removed. The change is...

However, when previously it was possible to revert utf-8 strings using negative range steps, it does only work with ascii-7 strings now...

Legitimate like, you use an invalid utf8 char on purpose to make a split between 2 strings, in a protocol. Well while it fix somehow, the fact that we have...

Rethinking at it, maybe we should provide the old behavior unchanged, and provide the new one via `StringByteSequence` operator `[_]`. Previous one was allowing to invert letters of the string,...

Argg we are doomed, `StringByteSequence[_]` returns a `Num` ... Have to sleep on this, to see if an idea pops up.

That does not work, `StringByteSequence` operator `[_]` Does not work with range because smart index returns the byte as `Num`. Yup, this is a mess from the start, I don't...

The thing is that there is absolutely no guarantee that any input even stdin is always utf-8, as per file redirection or broken writer... As per original byte container definition,...

That would also means we should remove the smart decoding of utf-8 chars, from String[Num]. We would be left with String as a glorified byte array, and have some sort...

I think we can reduce the problem by giving a max size to `readLine()`, so in code it should be something like: ```javascript readLine() { readLine(1024) } readLine(maxSize) { ......

Currently, modules are loaded from a "wren_modules" directory from the first parent directory, relative to PWD or script path (I don't remember exactly).