Fix in https://bitbucket.org/bobf/bpython/pull-request/58/fix-293/diff

Original Comment By:

The only one I personally encountered was sys.stdin.encoding. I think it makes sense to hold off on the unused stuff. Thanks.

Original Comment By: Steven Myint

So far I'm looking at adding:

from cli.FakeStdin

• isatty (return True)
• fileno (return 0)

from vanilla REPL:

• close (sets closed)
• closed (bool that does nothing)
• encoding (hardcoded to "UTF8" for now)
• name ('')
• sys.stdin.newlines (None)
• seek (nop)
• tell (nop)
• truncate (nop)
• writelines (nop)
• xreadlines (just alias to readlines for now)

However, it seems silly to write a bunch of code which may or may not do the right thing like this - I'd rather wait to hear that we need to implement something as a workaround. If our FakeStdin doesn't do something, it shouldn't implement it unless faking it makes things better for users. @myint, interested to hear what you and others think.

I don't know what to do with errors, but apparently need to find out.

I plan to add read later.

Original Comment By:

@myint I'm looking at all of the attributes of sys.stdin in a vanilla repl and wondering if I should really implement all of them. Do you need this for something, or was this just prospective?

Original Comment By:

Hm, we should probably add all of these then. Separate issue this example touches on: bpython-curtsies is currently hardcoded for utf-8. Is this a problem?

Original Comment By:

This looks like a bug that is specific to bpython/curtsiesfrontend/repl.py. Note that curtsiesfrontend.repl.FakeStdin is missing various attributes (and functions) that exists in cli.FakeStdin.

Original Comment By: Steven Myint

So you guys are moving to GitHub? By the way, the comments look quite out of order. Compare this to the original.

Oof, you're right. @ikanobori think it's worth fixing? We don't have tons of long conversations that evolve over time, but it's always sad to lose history.