CURLWS_CONT bugfix #16512
CURLWS_CONT bugfix #16512
Conversation
Will get a test going |
|
Analysis of PR #16512 at 780fc073: Test 2048 failed, which has NOT been flaky recently, so there could be a real issue in this PR. Generated by Testclutch |
|
I'm not sure what's going on, it seems like after I added my test case, every time the CI runs a different test fails, and it seems unrelated to my code? Could I get some advice here? |
roberte777
force-pushed
the
curlws-cont-bugfix
branch
from
108a6eb to
96f2b26
Compare
|
The test harness is slightly flaky when run in parallel mode (which is what the CI does) and almost exclusively on Windows which none of the main curl devs use, so it has remained unfixed. But, since parallel mode massively decreases the time to run tests we put up with it. Test Clutch will let you know if a particular failing test is known to be flaky but since it's not an issue with a specific test but rather something in the harness, you will seldom get such a notification. If you see a random failure in a single CI build that doesn't look related to your change, especially if it's on Windows (or sometimes pytest), then this is likely the reason. |
roberte777
force-pushed
the
curlws-cont-bugfix
branch
2 times, most recently
from
d3dccfd to
8335975
Compare
|
"This branch cannot be rebased due to conflicts" Can you please rebase this on top of master and force-push? |
roberte777
force-pushed
the
curlws-cont-bugfix
branch
from
0c45370 to
f31c9b1
Compare
Anything else I need to take care of before merge? |
|
Nope, we're all set for merge now! |
|
Thanks! |
Currently, CURLWS_CONT does not function as specified by the docs. This is an attempt to modify its behavior to be more inline with documentation.
This is my second attempt and attempts to implement the solution a drastically different way that will not change the existing API and instead tries to bring the functionality in line with the documentation.