I'll have a look. Want to do some tests.

First of all, thank you very much @jgebal for identifying the problem causing wrong warnings and providing this fix.

I've installed the PR and it works.

However, I've preferred the old sequence based solution. The timestamp based solution has the risk that the same savepoint name could be generated and used. Right now this seems to be irrelevant, since utPLSQL is running sequentially and there is some processing before requesting another savepoint name. But theoretically it all depends on the speed of the database and the granularity of the timestamp provided by the OS (microseconds or nanoseconds). - I do not like to rely on slow processing ;-) - So personally I'd like to include a fix in a way I've outlined in #982 (comment) . But I can live with the way it is done right now.

Thanks!