Joseph Murphy

And it's not a use case. It's a fundamental feature of being a Jupyter kernel. (that also happens to be my use case.)

Can Pexpect be modified to distinguish between the two (at least optionally)?

And I guess the problem is that Metakernel at the moment can't even distinguish if there was an error because it's using Pexpect that presently doesn't give information about which...

hmmm, but Popen processes can have their streams distinguished iirc. Of course you might not be remotely based on it, but how do they do it?

https://docs.python.org/3/library/subprocess.html#subprocess.CompletedProcess.stderr

So this may be a python core issue (assuming pty is an absolute dependency) but there shouldn't be a super good reason something generically communicating with another process shouldn't expose...

/ the cpython pty module is like 170 lines long. It wouldn't be incredibly difficult for someone (that someone might be me today) modify it and make a new module,...

Hi! Bump on that question of whether or not the stack could pull support down from a hypothetical segregated fd pty module.

200 lines of code presently. It's unclear how much of Pexpect would need refactoring. The options I put in are entirely optional arguments, but to actually use segregated streams would...

The pty core module itself is relatively tiny at 170 lines total because it is at could be described as some convenience functions that just garble the stdin, stdout, and...