From: Brian Candler <b.candler@...>
Date: 2010-05-11T18:10:28+09:00
Subject: Re: puts bug?

Roger Pack wrote:
> With this file:
> 
> # coding: UTF-8
> puts "asdfМикимаус"
> 
> output in windows is:
> 
> C:\dev\digitalarchive_trunk>ruby -v go.rb
> ruby 1.9.3dev (2010-05-11 trunk 27724) [i386-mingw32]
> asdfМикимаус
> 
> 
> Is this expected?

Yes.

One of the (many arbitrary) rules of ruby 1.9 encodings is that although 
there is an external_encoding set on STDIN, there is no external 
encoding set on STDOUT.

So when you write a string to STDOUT, by default the sequence of bytes 
is transferred as-is to the terminal.

If you want strings to be transcoded on write, try putting this at the 
start of your program:

STDOUT.set_encoding Encoding.locale_charmap

or

STDOUT.set_encoding Encoding.default_external

or

STDOUT.set_encoding "IBM437"

> Also shouldn't the following code fail?
> 
> "asdfМикимаус".force_encoding "IBM437"
> 
> (it doesn't)

No, it shouldn't; you're just tagging the string with a different 
encoding. In any case, notice that

  str = "asdfМикимаус".force_encoding "IBM437"
  str.valid_encoding?

returns true. This means that the sequence of bytes in str happens to be 
a valid sequence of characters in IBM437 (just not the characters you 
originally thought of)

However, the following does give an error, because you're trying to 
transcode each character in the source string to an equivalent character 
in IBM437:

  "asdfМикимаус".encode "IBM437"
  # Encoding::UndefinedConversionError: U+041C from UTF-8 to IBM437

If you want to understand how String encodings work in ruby 1.9 - and I 
warn you in advance that you probably don't - you could try reading
http://github.com/candlerb/string19/blob/master/string19.rb

Then cry.

Regards,

Brian.
-- 
Posted via http://www.ruby-forum.com/.