@cdybenko@uhlexsis FYI this is a common request from customers that I speak with and is related to your current work on thinking through relationships and hierarchies for work items.
@cdybenko@gweaver I've worked with dependency graph implementations in my past monitoring OSS project, and how to prevent circular loops or performance issues - may need query caching, or pre-calculated tree leaves. Happy to help with brainstorming ideas and design.
Naming proposal: {Issue,MR,Epic} Link Graph similar to the Git commit graph in Repository > Graph. Link because it summarizes relations and blocks described in the docs.
Another less linear example from Phabricator can be taken from https://phabricator.wikimedia.org/T214557, which highlights how and where this is useful in displaying task dependencies of large epics. (Not to conflate with #223035)
Please also take this paper's investigation and conclusion below as a suggestion to try out multiple ways of visualising ticket dependency in the long run:
Selectable and searchable text:
Our findings suggest that matrix layouts are the best layouts to investigate
direct causal relationships, with matrix-out-degree the fastest, while node-link
diagrams with hierarchical layout is the most promising for mediated causation.
For identifying causal structures, radial was the most promising layout, with
its performance increasing with the size of graphs. This suggests that causal
inference could benefit from visualisation tools that provide multiple coordinated
views [26], thus supporting users in a range of different tasks for understanding
causation. Further investigation that considers cognitive and visual processes
would help in explaining the results of our experiment, and better understanding
of the principles of visual causal inference will assist in developing readable and
informative causal graphs.