DAG dependency tree manual build flows
Hi,
I'm experimenting with adding a new page to the dashboard and adding a DAG (directional acyclic graph) flow to manually build the dependency job tree; i.e. with wait barrier where edges come in to a node/job and kicking off nodes/jobs down stream if successful.
I'm using http://js.cytoscape.org/ ; would such a dependency be acceptable provided the application of it reaches your standard?
If there is potential interest in this, I'll fork and show my work as I go
Can you explain what the goal is? It sounds more of a feature for https://github.com/Strider-CD/core
Exploring addition of a directed acyclic graph flow both visually and programmatically. It allows the user to connect the edges, the job to job dependencies, start a directional build manually. And monitor progress and status interactively(possibly graph animation and tips). There would be a model(e.g. https://github.com/jsongraph/json-graph-specification) of the flow/pipeline sent back to mongodb to persist the state or be shared between users. It would build jobs downstream conditionally by clicking a node/job and have the latest changes from artifact copies from upstream successes. State would be illustrated by color. And the flow could be triggered by other events such as cron as well. The problem fits graph theory and cytoscape.
Currently an example looks like

If there is interest I'd appreciate discussion before I go too far without a solid set of features people would want; use t strider technology correctly(maybe flows could be built as plugins?). Just thinking from my own projects perspective up till now. Many have c++ and native dependencies. I want to use the speed and efficient artifacts in strider too
Related issue here: https://github.com/Strider-CD/drone/issues/1 We discussed using noflo (which is graph based and can run any code)
Thank you, I'll read and learn