Erik Hennum

Thanks for isolating a potential hotspot. One concern -- as I understand it, DOM implementations are not required to be thread safe. I'm wondering whether it might be best to...

@maffelbaffel , thanks for the thoughtfulness invested in this pull request. A belated question occurred to me: did your performance testing indicate how much of the DOM initialization cost is...

I should have been more clear -- I was wondering whether it would make sense to cache DOMImplementationLS instead of DocumentBuilder because both the LSParser and LSSerializer are created from...

In comparing the first and second flame graph, I didn't see a big difference in the overall profile. However, I'm not a performance engineer specialist. As usual, the engineering concern...

I think I see what you're saying. The DOM implementation is strange in that the application pays for initializing the DocumentBuilder object (which is itself unsafe) even though the DOMImplementationLS...

@ling, it's surprising to me that the call needs to be wrapped. As I read it: - bucketGroup() returns a PlanNamedGroup. - PlanNamedGroup extends PlanNamedGroupSeq. - So, the return value...

+1 both that the fix may be on the server side and that the availability of the workaround makes it possible to push this bug.

The root issue is that OkHttpClient.Builder.sslSocketFactory() is deprecated in Java 8 and throws a runtime exception in Java 9 because a X509TrustManager is required. At the next backward incompatibility boundary...

@georgeajit , commit c13902c862d5285bfcac8524621fcb61d166a03c appears to be in the commit log for the develop branch. I also notice that GitHub associates it with the 4.2.0 tag. Finally, it looks like...

A couple of notes. First, the following tests are affected: testRawCtsQuery testRawValueQuery testRawCombinedQuery testDirectoryQuery testCollectionQuery testStructuredQuery testStringQuery Second, if develop-143 proves beneficial and thus is merged, this bug will be...