Iterating over a byte file object using __iter__ is rather useless, and a bit confusing perhaps.

It'd be nice to propose two things:

1. Having __iter__ raise an exception if the file was opened with the "b" flag.

2. Adding a new by_bytes() method to file objects, which would return an iterator that yielded the next byte in the file, introducing the ability to write a similar `for byte in byte_file_obj.by_bytes()` that is currently possible when iterating by line over non-byte files.

It's quite easy to do (2) currently, obviously, but perhaps it's worthy of consideration for inclusion.

FWIW, an idiom I use in Py2.x is:

   for block in iter(partial(f.read, BLKSIZ), ''):
        . . .

This works with both single bytes at time and multiple bytes at a time.

BufferedIOBase readline and __iter__ has to stay for backwards compatibility (especially with python 2). As for by_bytes(), I suggest you post it to python-ideas.

For the record, I don't find the behavior of __iter__ on a binary file at all confusing.  It's the same behavior I see if I open the file in, say, vi.  So it is in fact the behavior I expect, and I would be surprised if it didn't work.

Whether it is *useful* is a different story, but I'd be willing to bet that there are or will be python3 programs that do use it.