cbaWorkflow : design and implementation for node levels start

generating adjacency matrix from nodes_cadence table

Adjacency matrix represents all the possible pathways in the execution graph.

Execution graph is  a directed acyclic graph as all the nodes dependencies are only resolved from nodefrom to nodeto

There are no loops in traversing the graph.

the distinct nodes values are read and stored in a  vector<string> nodes

a square matrix (vector<vector<int>>) adjacency_matrix is created with the size of the nodes vector and initialized to 0

using the values of nodefrom the vector<int> is populated with value of 1 for each nodeto in a loop to populate adjacency_matrix


generating levels for the execution start

Level 1 is the vector[0] of the adjacency_matrix, all the nodes having a value greater then 0 are started

Level 2 matrix is created using dot product of adjacency_matrix  by itself, level2 is the the vector[0] of the level 2 matrix

Level 3 matrix is created using dot product of adjacency_matrix by Level2 matrix, level 3 is the vector[0] of the level 3 matrix

Level N matrix is created using dot product of adjacency_matrix by Level(N-1) matrix, level N is the vector[0] of the level N matrix

when Level N vector has only one value for the end node, generation stops

Level 1 and Level 2 nodes are started at the beginning of the run

Level 3 nodes are started when at least one Level 1 node completes

Level N nodes are started when at least one Level (N-2) node completes


creating library functions for matrix and level vectors creation

The matrix generation will be implemented in cbaUtils class

The matrix dot product will be implemented in cbaUtils class

the Level  vectors generation will be implemented in cbaUtils class



Attachments: