Hi Thibault, long time no see!

So is the purpose of this PR to provide a basis for a general mitigation strategy that shall be implemented independently, or do you intend to ultimately propose a PR that could be merged?

Either way, thanks for trying out Serena and for pointing out the issue.

Hi Dominik (I was surprised when I realized this project is by you),

my first intent was to provide an immediate github fork of Serena that can be installed and used by others in my company, and to get feedback.

In the medium term of course I would like to delete my fork again once we found a way to run the JDTLS by Serena with gradle projects in my company, e.g. by merging changes such as those in this MR (or a better fix).

At the least I think we would need a way to configure the gradlew wrapper and JRE behavior, and this MR could show a way, though I am not sure whether the MR is good/minimal, I did not have time to polish what the AI produced.
But I am unclear on why gradle is called at all, and with what tasks. Potentially there is another mitigation whether the call is not made or made differently.