[erlang-questions] Different behaviour of the erl shell and compile:file

[Aleksandr Vinokurov]

Hello all,

I'm investigating Unicode support in Erlang. And have found the difference
in behaviour of erl shell and compile:file.

The erl shell:

1. We configure our terminal emulator to use UTF-8
2. We set io:setopts({encoding,unicode}).
3. Then in the prompt we enter a Unicode string and see the result -- a
list of Unicode codepoints:
(emacs@REDACTED)103> "АБВГД".
[1040,1041,1042,1043,1044]


The compile:file:

1. We save a file foo in UTF-8 with a function that return a Unicode string
bar() -> "АБВГД".
2. Now we open an erl shell and run compile:file(foo).
3. Finally running foo:bar() we have a list of UTF-8 bytes.
(emacs@REDACTED)111> foo:bar().
[208,144,208,145,208,146,208,147,208,148]


I dug the compile:file/1 and found epp:server/4 where the file is opened
with file:open(Name, [read]) and where encoding of the returned IoDevice is
not configured.

Why not to use one of the Unicode recipes from the Unicode usage doc
(lib/stdlib-1.18/doc/html/unicode_usage.html) to set
io:setopts(F,[{encoding,Type}]) or add an encoding option to the
compile:file/2 and erlc?

What would be your suggestions, honorable all?

-- 
Александр Винокуров
+7 (921) 982-21-43
@aleksandrvin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20120319/c5982c9a/attachment.htm>