Hi Mihal,
When any code can be told to be 100% error free? By definition, never. In most of the cases ETSI doesn't validate the test code, but expects members & industry to report errors/probelms/findings. This is also the case here. The code is using a tool extension of a TTCN-3 tool and it has not been reported to ETSI yet (at least up to the version you are using). These are the facts of life we have to leave with.
So, the way forward is that if you are not using TTWorkbench, you correct the *-s to ?-s in your code (as in other tools, if it at all compiles, the runtime semantics is not defined) - and report the problem to ETSI in Mantis:
forge.etsi.org/mantis/main_page.php, project ETSI TTCN-3 Libraries/SIP Library (the issue may also be present in the test case code, not (only) in the library part, but I cannot fing a better place at the moment).