homepage

This patch causes the use of builtin file to give a Py3k warning.
When Python starts up, distutils.text_file gives a warning because it
uses a variable named file. I imagine there are places like this all
over in the stdlib, which should be renamed fp or something similar.

I'll review this. My hunch is that we don't need this -- 2to3 takes
care of this so there is no reason to tell people to change their code
by hand. 
(You may notice a pattern -- things that 2to3 can fix easily generally
don't deserve -3 warnings or backports, *unless* the backported feature
adds significant functionality that's not easily available otherwise in
2.x.)

Let's not do this. This approach is not sufficiently backwards
compatible; it will break any code that uses isinstance(x, file). Even
though that's not forward compatible with 3.0, it works in 2.5 and
before, so it should not break in 2.6. Together with my previous remark
this means that we should just not mess with this.
(PS: I don't understand what you say about distutils.text_file -- from
your patch it looks like the only way it can issue this warning is if it
actually calls the file() builtin...)

Yeah, breaking isinstance(x, file) would be bad and 2to3 can do this
automatically. I should probably learn how to write fixers. BTW, the
warning culprit was actually site.py which used file.