Markus KARG

@lukasj Does it makes sense to set up a PR for factory caching or do you see architectural constrains that this would violate?

Thank you for picking up this thread. As nobody responded in the past months I already have removed the benchmark and PR. But as the problem persists, I will invest...

No need to. I have already finished the PR tofay. Please just stay tuned. Holding back just for more testing this week.

@laurentschoelens Please have a look at this draft: https://github.com/eclipse-ee4j/jaxb-ri/pull/1855. I assume that is what you had in mind, too. It passes all functional tests (so it does not *break* something),...

> Could you tell me the command for creating one and the place where it is located? 🤔 Answering myself: Found the runtime JAR in `runtime/impl/target`. Seems I missed the...

I do not see that anybody actually would call a public method from an `impl` (internal) method, even when it is public. And *if* he gets a cached instance, that...

Asking again: Shall I extend my PR to prevent breaking external things (which simply means to introduce a non-public method), or do you still want to put a PR *ontop*...

See [Original Thread in SAXON Forum]( https://saxonica.plan.io/boards/3/topics/9475?r=9564#message-9564) for a deeper understanding of the scenario. JAXB-RI simply needs "forever" (= several seconds) for unmarshalling because: * `JAXBContextImpl` creates a **new** `TransformerFactory`...

> > In complex scenarios (we do have hundreds of JAXB-annotated classes in one context and tens of thousands of lines of XML), JAXB-RI is calling createTransformer / createTransformerHandler a...

Had no time in the past weeks, will deliver ASAP.