fat stack (fixed)
I checked in yesterday an important modification in the heart of OpenWFEru. It makes this open source ruby workflow / bpm engine way less ‘beta’.
The initial, naive, implementation of the process execution of OpenWFEru could generate deep stacks. For small processes, it was OK, but it broke at a certain point (like, for instance, a sequence of 450 participants), and it got slower as the stack of a process grew.
Processes with asynchronous participants (outside of the Ruby interpreter running the engine) were less susceptible to that problem.
This has been fixed. One of the consequences of the fix is that the launch() method lost its optional ‘async’ parameter. Now (OpenWFEru 0.9.10), all launches are asynchronous. Don’t argue, it’s a business process execution engine, it deals with time and concurrence. (You can still wait_for() a given process instance).