Conversation
|
@dataflake It seems we still have cache poisoning in Windows PyPy: I deleted the cache to get #127 green and merged. (I will not start working on it just now.) |
|
Observation just now, unsure if this is relevant:
|
I do not fully understand how restore-keys works, but I have seen a definite cache collision where the restore-keys value `Windows-X64-pip-` led to a `Cache hit for restore-key: Windows-X64-pip-3.14` for the original cache key `Windows-X64-pip-pypy-3.11`
|
After going through the raw logs I saw instances where the The I do not know if the attempt to provide a fallback cache key provides any value, after removing all instances in 03078a8 things didn't seem to run slower. But the actions run succeeded immediately. I know this |
+1. An incorrect cache is worse than no cache. |
Created using zopefoundation/meta#416