[Python-3000] New io system and binary data

[Guido van Rossum]

On 9/19/07, Fred Drake <fdrake at acm.org> wrote:
> On Sep 19, 2007, at 3:58 PM, Jason Orendorff wrote:
> > Given the context (stdin/stdout/stderr), I'd love to know what you're
> > thinking of here.  I can't name a program offhand that wants to
> > operate on binary data via a pipeline.  There are a few that *can*,
> > like gzip, but my impression is that even those aren't often used that
> > way anymore.
>
> Huh.  I use pipelines constructed in the shell for binary data
> regularly; I don't see any reason not to do that.  I'd certainly
> rather see the stdio streams be available as binary data, possibly
> with convenient text-centric wrappers also available.  But I'd be
> fine with constructing those myself.

I agree that binary pipelines are useful and should be possible. I
just don't think this should be the default behavior for stdin/stdout.

Since the binary stream underlying stdin is readily available as
sys.stdin.buffer (and ditto for stdout and even stderr) I don't think
any action needs to be taken. note that the instance variable doesn't
start with an underscore. It's part of the public API for text files.

-- 
--Guido van Rossum (home page: http://www.python.org/~guido/)