[erlang-questions] Atom Unicode Support

Jesper Louis Andersen <>
Tue Feb 2 13:46:56 CET 2016


On Tue, Feb 2, 2016 at 1:26 PM, Fred Hebert <> wrote:

> I want to also voice my support for Unicode support.
>
> I don't know why everytime unicode is brought up in this mailing list a
> bunch of people suddenly fear having to edit code in a language they don't
> understand.  This has been possible already for a long time (as pointed out
> by the list of languages José added there). I think last time someone was
> being preemptively angry because they could be buying a business where code
> was in a different language and then they would be screwed! The horror.
>

I'm in favor of Unicode support for atoms as well as a generic
construction. Squint your eyes enough and an atom() is merely a string
literal of a specific kind. It makes sense to allow these to vary inside
the Unicode world.

When used as an "identifier" it may be necessary to define certain
restrictions as to what symbol is allowed. The Go language defines that an
identifier must start with either '_' or a Unicode symbol from the Letter
class in order to avoid certain notational headaches such as the symbol
'123' which cannot be written as 123 in Erlang either. In go, you could
write _123, but that will get through the erlang parser: it is a variable
since it starts with the wildcard character.

There are some places where notation can help a lot. When transcribing math
algorithms, it is often more readable to use the symbols used in the
mathematical text directly in the source code as binding values. And the
above remarks also suggest it is better to give the programmer the freedom
to choose here, rather than restricting what the user can do.

Syntax is nothing, semantics is everything :)

-- 
J.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20160202/67953ad3/attachment.html>


More information about the erlang-questions mailing list