vlsi commented on PR #113:
URL: https://github.com/apache/xalan-java/pull/113#issuecomment-1808387594

   > If you happen to know of a trustworthy implementation that could be used 
from a regression test, I'd appreciate a pointer
   
   * 
https://github.com/mjiderhamn/classloader-leak-prevention/tree/master/classloader-leak-test-framework
 can ensure classes can be unloaded
   * https://github.com/openjdk/jol can inspect into object contents. For 
instance, `GraphLayout.parseInstance(...)`: 
https://github.com/openjdk/jol/blob/9d62009e97a1756ae91a16ab9e72f294efce17a9/jol-samples/src/main/java/org/openjdk/jol/samples/JOLSample_24_Arrays.java#L75C49-L75C49
 or use java heap dump as an input. One can collect heap dump with 
`com.sun.management.HotSpotDiagnosticMXBean#dumpHeap`
   
   > Especially in hotspot JIT compilers, which can optimize differently every 
run. Again, specific proposals for reliable automatable metrics that we could 
add to our toolkit are welcome
   
   https://github.com/openjdk/jmh is the way to go.
   
   For instance, see 
https://github.com/pgjdbc/pgjdbc/blob/b49b61806f050ed13a255ad35449774297fad8c2/benchmarks/src/jmh/java/org/postgresql/benchmark/escaping/EscapeProcessing.java#L49-L53
   
   > it's just nulling one field
   
   I might be wrong here, however, as `m_readers` becomes `null`, the 
subsequent calls to `getXMLReader()` become slower since they would have to 
create a brand new reader instead of reusing the previously cached one in the 
`ThreadLocal`.
   
   In other words, I believe this PR makes `ThreadLocal m_readers` useless. If 
the caching was not needed in the first place, then `ThreadLocal m_readers` 
should be removed. If the caching was needed, then this PR introduces 
performance regression.
   Of course, there's an option I am wrong, however, well, I am quite sure my 
analysis is right.


-- 
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.

To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org

Reply via email to