Dominik Helm

I don't think the potential solution is as bad as you think it is. Yes, there are many uses of that constructor, but it seems that a) many are parts...

Thanks for your investigation! For RTA: Question is, do we want this. These types are instantiated by the JDK, but on other system, e.g., on Android, they might not be....

I think we should deal with this in OPAL, we should just be sure where to properly deal with it. I suppose at least for the types that we already...

Oh, now I understand your suggestion. I don't think we can just use any subtype of the field's type even if that type is not in the ExternalWorld set, since...

Thank you for the extensive investigation and the PR. I fully agree with the additions to the config and with adding instantiated types to `ExternalWorld`. I'm a little wary however...

This solution sounds reasonable. Thank you for your work! About the `ExternalWorld`: true for _manual_ specification of types. A suitable `InitialInstantiatedTypesFinder` would do this automatically however.

You can have a look at `ConfiguredMethodsPointsToAnalysis`, it needs to do the same for allocation sites in native or otherwise unavailable methods. The main reason for choosing the second option...

Shouldn't the second point already be addressed by the configured native methods analyses? I thought it was.

You are right, they are configured to use the RTA class there, which doesn't seem to make much sense. Not sure why that is, I thought there already was a...

As the failed it:test shows, this PR needs re-generation of the `FPCFAnalysesIntegrationTest` reference files. This seems to be because of changes to the entities for the escape analysis? Given the...