| 1 |
This directory contains test cases that are known to leak references. |
| 2 |
The idea is that you can import these modules while in the interpreter |
| 3 |
and call the leak function repeatedly. This will only be helpful if |
| 4 |
the interpreter was built in debug mode. If the total ref count |
| 5 |
doesn't increase, the bug has been fixed and the file should be removed |
| 6 |
from the repository. |
| 7 |
|
| 8 |
Note: be careful to check for cyclic garbage. Sometimes it may be helpful |
| 9 |
to define the leak function like: |
| 10 |
|
| 11 |
def leak(): |
| 12 |
def inner_leak(): |
| 13 |
# this is the function that leaks, but also creates cycles |
| 14 |
inner_leak() |
| 15 |
gc.collect() ; gc.collect() ; gc.collect() |
| 16 |
|
| 17 |
Here's an example interpreter session for test_gestalt which still leaks: |
| 18 |
|
| 19 |
>>> from test.leakers.test_gestalt import leak |
| 20 |
[24275 refs] |
| 21 |
>>> leak() |
| 22 |
[28936 refs] |
| 23 |
>>> leak() |
| 24 |
[28938 refs] |
| 25 |
>>> leak() |
| 26 |
[28940 refs] |
| 27 |
>>> |
| 28 |
|
| 29 |
Once the leak is fixed, the test case should be moved into an appropriate |
| 30 |
test (even if it was originally from the test suite). This ensures the |
| 31 |
regression doesn't happen again. And if it does, it should be easier |
| 32 |
to track down. |