aenachescu / nbp_old Goto Github PK

I created this to keep notes. I will create a branch too.
Workflow:

• Create test structures in NBP_CALL_TEST or NBP_CALL_TEST_CTX storing rules and their names
• Create a graph structure containing one node for each test (the dependency graph)
• Based on the number of tests (and maybe other configuration) create a large enough hashmap to achieve a good enough performance
• Store all test names and their respective structure pointer in the hashmap
• For each test, iterate over rules and resolve their targets (ie. get test pointer by name from the hashmap)
• For each rule, create an suitable edge in the dependency graph
• Detect if there are any cycles in the graph
• According to the configuration, create N threads and an execution queue. Fill the queue with the tests that have no dependencies
• Execution flow should mimic a BFS algorithm. A test node should be added to the queue only when all its dependencies are met.