Remove relation edge attribute and reporter edges#63
Conversation
| start_task = function(task) { | ||
| task_graph_package_status(self$graph, task) <- STATUS$`in progress` | ||
| start_node = function(node) { | ||
| task_graph_package_status(self$graph, node) <- STATUS$`in progress` | ||
| private$start_node_meta_parents(node) | ||
| }, | ||
|
|
||
| start_node_meta_parents = function(node) { | ||
| # start parent meta tasks recursively once a child starts | ||
| parent_nodes <- igraph::adjacent_vertices(self$graph, node, mode = "in") | ||
| parent_nodes <- unlist(parent_nodes, use.names = FALSE) | ||
| parent_nodes <- V(self$graph)[parent_nodes] | ||
| meta_parent_nodes <- parent_nodes[is_meta(parent_nodes$task)] | ||
| for (meta in meta_parent_nodes) { | ||
| task_graph_package_status(self$graph, meta) <- STATUS$`in progress` | ||
| private$start_node_meta_parents(meta) | ||
| } | ||
| }, |
There was a problem hiding this comment.
This is probably the most important change. When new processes are started or finished, the status of parent meta tasks are also updated (recursively).
If a meta task has any children that are active, it will be active. Once all children are finished, it also finishes.
This made it pretty straightforward to get everything working with the reporter without the relation edges attributes.
| push_process = function(task, x) { | ||
| task_graph_task_process(self$graph, task) <- x | ||
| name <- task_graph_task_name(self$graph, task) | ||
| private$start_task(task) | ||
| push_process = function(node, x) { | ||
| task_graph_task_process(self$graph, node) <- x | ||
| name <- task_graph_task_name(self$graph, node) | ||
| private$start_node(node) |
There was a problem hiding this comment.
More of a refactor than a substantive change, but I found this hard to reason about because task is actually a vertex. I felt like this would make it a bit easier to develop against.
maksymiuks
left a comment
maksymiuks
left a comment
There was a problem hiding this comment.
Looks cool! It will definitely simplify the graph and make it clearer. Thank you for that!
2 participants
Will highlight specific considerations in comments