Fix tests on Windows, add Windows CI #1186
Conversation
| @@ -2524,7 +2524,7 @@ def check_recovery( | |||
| b"-passin", | |||
| b"pass:" + passwd, | |||
| *extra, | |||
| ) | |||
| ).replace(b"\r\n", b"\n") | |||
There was a problem hiding this comment.
_runopenssl is also used with binary output, so we can't replace this in there for all invocations.
| @@ -2838,23 +2845,24 @@ def test_wantWriteError(self): | |||
| """ | |||
| client_socket, server_socket = socket_pair() | |||
| # Fill up the client's send buffer so Connection won't be able to write | |||
| # anything. Only write a single byte at a time so we can be sure we | |||
| # anything. Start by sending larger chunks (Windows Socket I/O is slow) | |||
| # and continue by writing a single byte at a time so we can be sure we | |||
There was a problem hiding this comment.
This brings this test down from 16s to <20ms on my machine.
There was a problem hiding this comment.
Socket I/O performance joins VFS performance on my Windows is bad list 😄
|
The Windows CI job fails to upload coverage right now. Do we care about this? I do maintain https://github.com/mhils/better-codecov-action, but I don't want to impose my homegrown actions on anyone else. 😛 |
|
Haha, well you're doing the work. We could also adopt the coverage system cryptography uses (which eliminates codecov entirely), although it'd probably need to be updated to report back via a status job rather than just failing the build when it's under 100% since we don't have 100% here :) Either way let's take it as a follow up PR after this if you want to tackle it. |
No description provided.