synchronized block -> getCrashInfo -> join -> synchronized on the same
context on the child thread -> deadlock
instead we check for isTerminated and then get the crash info outside
of the synchronized block.
closes#6669
this happens because isTerminated returns true before the thread's shutdown handler runs,
so we join with the thread to make sure that shutdown handlers are done before returning.
... hopefully we don't get servers randomly deadlocking in shutdown handlers ???
ErrorToExceptionHandler currently prevents these from appearing by turning them into exceptions, but this won't always be the case.
For example, in the future we may not want to turn ALL types of E_* errors into exceptions (e.g. E_DEPRECATED).
there's no need for the worker specialization here (isShutdown and shutdown are aliased to isJoined and join respectively), and the unstacking is not really desirable either as we previously learned with AsyncPool.
Despite the comments, there doesn't seem to be an obvious reason for these to be copy-pasted. Perhaps there was some legacy reason for this with legacy pthreads.
In fact, it looks likely that quit() will probably be able to be traitified too.
until now, any thread crash would show as a generic crash since we aren't able to get the trace from the crashed thread directly. This uses some dirty tricks to export a partially serialized stack trace to the main thread, where it can be written into a crashdump.
This enables us to see proper crash information for async tasks in the crash archive (finally!!!) as well as being able to capture RakLib errors properly.
the only use for this class is to facilitate random runtime plugin loading, and it's not complete even for that purpose.
Since nothing but PM uses pocketmine/classloader anyway, it doesn't make sense to have it outside the core. As with LogPthreads, it's just adding more maintenance work.
this was previously part of the abandoned package pocketmine/spl. It had to be separated in the PM3 days, because RakLib depended on it.
Since RakLib 0.13, RakLib stopped being dependent on or aware of pthreads, so it no longer depends on any thread-related packages.
It's also possible to absorb pocketmine/snooze and pocketmine/classloader back into the core with this in mind.
this prevents any tasks still left in the queue on shutdown getting pulled out by the worker when we're attempting to shut it down.
This led to various race conditions, most notably weird cases where PopulationTask would inexplicably find its expected generator state had not been correctly set up.
this has the most reliable behaviour, because some stuff like respect/validation likes to include files unconditionally which declare functions that are already declared, which breaks due to the lack of function autoloading.
