Currently, OptimizeInputsTask avoids a dead-loop with a sequential traversal of a potentially long list of tasks, with a dynamic cast for each of them (see #29). This alternative uses a hashset of previously visited expressions for the same purpose, similarly to explored_group.
Currently, OptimizeInputsTask avoids a dead-loop with a sequential traversal of a potentially long list of tasks, with a dynamic cast for each of them (see #29). This alternative uses a hashset of previously visited expressions for the same purpose, similarly to explored_group.
This is the second attempt, after #196.