Fix some typos/missing words #81
Conversation
Codecov Report
@@ Coverage Diff @@
## master #81 +/- ##
==========================================
- Coverage 98.29% 98.24% -0.06%
==========================================
Files 50 50
Lines 5876 5876
Branches 464 464
==========================================
- Hits 5776 5773 -3
- Misses 86 88 +2
- Partials 14 15 +1
Continue to review full report at Codecov.
|
|
Wow codecov is so confused! |
|
@dimaqq actually it's accurate since I did add lines to the project that didn't increase the coverage. I believe there's a way to say "ignore these directories" which would solve this. |
|
@njsmith FYI Codecov has a YAML file you can check in to configure it if you want help tweaking settings. |
|
No, codecov doesn't care about README -- the problem is that there's a branch in |
|
@njsmith have you thought about using |
|
It's a legitimate complaint though :-) We should have coverage for both branches, just I got lazy about writing tests for a few bits of code that aren't actually used yet, and are just placeholders for future expansion. (Specifically the non-socket APIs for Windows and Kevent, which I figure will get sorted out properly once we add subprocess support - see also #4, #26, #52.) I have a patch here that I think makes the test coverage deterministic at least though... |
There's a race condition when we signal the IOCP thread to exit by closing the IOCP: the thread might be blocked in GetQueuedCompletionStatusEx, or it might (rarely) be in the rest of the loop. The effect of this is that there are two possible error messages we might get that both indicate that the IOCP was closed. Before we were only detecting one of them; now we detect both. This has the side-effect of making it so that during the test suite we deterministically only take the regular-exit path, rather than sometimes taking the blow-up-on-unrecognized-error path. Which is good because codecov was getting cranky about that non-deterministic coverage and yelling at people for no reason (see python-triogh-81).
No description provided.