Repository navigation
Experimental debugger fails to stop at breakpoint on CI server (multi-threaded code) #1250
Description
One of the unit tests (Test multi-threaded debugging) fail almost -25%-50% of the time when using the new PTVSD debugger. What's interesting is the same tests pass 100% of the time when using the old PTVSD debugger.
Here's the python code
print("step1")
import sys
import threading
import time
print("step2")
def bar():
time.sleep(2)
print("print")
def foo(x):
while True:
bar() # add breakpoint here <-
threading.Thread(target=lambda: foo(2), name="foo").start()
foo(1)In the unit test I add a breakpoint to the line bar() inside the loop.
I can see (after placing print statements) that two threads are running the while loop.
However, the tests fail as the debugger fails to break at the break point.
Issue filed upstream microsoft/ptvsd#282
Activity
DonJayamanne commentedon Mar 30, 2018
Adding a few delays and delays seem to increase the success rate. But that's a work around, old PTVSD debugger didn't require any such changes.
Parallelize unit tests on CI server (Travis) (#1251)
Enable multi-threaded tests for experimental debugger (#1935)