michael-o edited a comment on pull request #77:
URL: https://github.com/apache/maven-resolver/pull/77#issuecomment-817367241
There is one more thing I do not understand and think that this undermines
reentrancy: While I understand that you maintain a concurrent map in
`NamedLockFactorySupport` to manage locks (counts) between threads in one JVM
to properly obtain and release them, but I fail to understand why both
`AdaptedReadWriteLockNamedLock` and `AdaptedSemaphoreNamedLock` use thread
locals to manage these locks. Looking at how lock and unlock are done, I don't
see how underlying locks are used when reentrancy happens. Thus, the internal
lock count is not increased, but you turn reentrancy into a `NOOP` and the end.
Each call to `lockShared()`/`lockExclusively()` should be delegated to the
underlying lock , no matter what. Please explain!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
