From: Bill Kelly <billk@...>
Date: 2010-02-10T06:38:15+09:00
Subject: Re: Strings -- was: Re: Why Ruby?

Marnen Laibow-Koser wrote:
> 
> A string is composed of characters.  Binary data is composed of bytes. 
> They're two different things and should be supported as such in the 
> language.

I was about to disagree.  But then I remembered I have some ruby 1.8
code like this:

  while line = gets
    line.length.times do |i|
      line[i] ^= 0x80 if (line[i] & 0x7f).between?(0x21, 0x7e)
    end
    print line
  end

...which while it's still technically operating on "text", it's
7-bit ASCII text where I need to toggle the 8th bit.


With 1.9, I guess this becomes something like:

  while line = gets
    line.force_encoding Encoding::BINARY
    line.length.times do |i|
      val = line.getbyte(i)
      if (val & 0x7f).between?(0x21, 0x7e)
        line.setbyte(i, val ^ 0x80)
      end
    end
    print line
  end

....except, I don't trust the 'gets' and the 'print' here.  If I
were doing I/O to a file I'd opened directly, I'd use "rb" and
"wb" modes to get ASCII-8BIT encoding.  But this is stdin/stdout.
So I can't expect the encoding to be binary-compatible.

Maybe before the loop, instead, setting:

  Encoding.default_external = Encoding::BINARY
  Encoding.default_internal = Encoding::BINARY

  while line = gets
    line.length.times do |i|
      val = line.getbyte(i)
      if (val & 0x7f).between?(0x21, 0x7e)
        line.setbyte(i, val ^ 0x80)
      end
    end
    print line
  end

...the above indeed appears to work.  (I presumably instead could
have specified -EBINARY:BINARY on the command line, but I didn't
want to have to remember that when running this tool.)



So..........

I guess this is a weird hybrid case, since I'm doing binary
processing on text.  :)

So even if Ruby had a data type for binary data that was separate
from String, I'm not sure that would help in my hybrid case where
I still want to use 'gets'.

:shrug:


Regards,

Bill