Improve support for later Python versions #13
Conversation
|
So this seems like doubling down on a wrong way. Maybe we should stop using RW wrappers regardless of version. And since there's output pre-processing anyway I'd just pass the socket instance to the LF2CRLF_FileWrapper - seems more clean and encapsulated. |
|
I agree, it does seem horrid to go further on this. That said, reliably getting lines out will also be a fair chunk of code so I don't think it will be simpler overall. An alternative would be simply to live with the bug, or make the tests resistant to it. It will only affect users who copy/paste or otherwise automate interacting with remote-pdb, normal commands will still work. There's certainly an argument for just documenting that some versions of Python may drop lines in rare circumstances. |
MatthewWilkes
force-pushed
the
python3-compatibility
branch
from
16fc7df to
393197d
Compare
|
Well I guess I need to write an integration test to see how bad this bug is. I'll get around to it ... uh ... I have a terrible flu/cold, soon-ish. |
when data is sent. This prevented multiple rapidly entered commands from all being received, causing failing tests and the inability to paste commands into the telnet window.
ionelmc
force-pushed
the
python3-compatibility
branch
from
393197d to
b565296
Compare
|
So I added a test + simpler change, lets see how it works on CI ... |
|
Sorry, I missed your last couple of messages, had a bit of a crazy time myself. Thanks so much for fixing this up, I really appreciate it! |
Hi!
This PR adds testing for newer Python versions, as well as a workaround for a CPython bug in Python 3.6.6 and up.
It works for me locally, but there are a lot of tox environments and I have only run a subset of them. I'll fix any errors that come up, but if you have any concerns please do holler.
Cheers, love the software,
Matt