Do I have these string definitions right?

[rvirding]

@Qqwy

Thank you very much for clearing those things up, @rvirding!

if I correctly understand how UTF-8 list strings are built, a four-byte UTF-8 encoded character will be represented in list string format as four separate elements as well. Or am I wrong?

Generally if you are using list strings each element of the list is one codepoint as there is no need to split them into multiple “bytes”.

Your other two arguments (copying measured as taking a long time, codepoint-access being linear for UTF-8 strings regardless of format) are indeed very true. I would love to see the work of the person who measured the performance, just to see in what cases one is better over the other.

IIRC this is the start of the thread:

http://erlang.org/pipermail/erlang-questions/2012-October/070067.html

[JEG2]

I think I finally understand why these definitions are confusing to me:

1. Elixir defines (or copies) a chardata type
2. That definition is dependent on a “charlist” type
3. That charlist is not Elixir’s charlist type (which is a renaming of Erlang’s “string” type)

Do I have this right?