Christian Bartolomäus

> It seems to work with a BEGIN in front of my Lock $l .= new. For the record: In Rakudo's ```lib/NativeCall.rakumod``` an ```INIT``` has been added to work around...

Are there any opinions about this? An alternative would be to special case this for the JVM backend. But from my understanding the extra check is logical.

I noticed that adding ```sleep 1``` after the call to ```hyper``` made the error message show up every time (and seemingly ```nqp::continuationcontrol``` always got called twice). So probably the "missing"...

Some observations: For some reasons ```GLOBAL``` seems to be ```null``` here: https://github.com/rakudo/rakudo/blob/37f5259958/src/core.c/stubs.pm6#L53 Sidestepping that problem there is another problem with ```$*ERR```: The call to ```$err.say``` in https://github.com/rakudo/rakudo/blob/37f5259958/src/core.c/Exception.pm6#L555 fails. (It complains...

I've found another test that fails in the same way: https://github.com/Raku/roast/blob/54762714e5/integration/error-reporting.t#L200-L203 ``` $ ./rakudo-j -e '#`{{ my long unfinished comment' Unhandled exception: java.lang.NullPointerException in DYNAMIC (gen/jvm/CORE.c.setting:849) in print_exception (gen/jvm/CORE.c.setting:53657) in...

I kind of rediscovered this bug today. If I'm not mistaken, ```GLOBAL``` is ```null``` at this place on MoarVM as well: https://github.com/rakudo/rakudo/blob/37f5259958/src/core.c/stubs.pm6#L53 But on MoarVM this doesn't blow up: ```...

I've created a PR with the above patch: https://github.com/rakudo/rakudo/pull/5379 Maybe it's not the right solution for the problem, but at least we could discuss this idea.