cisagov--redeye/packages/graph/TODO.md

GRAPH TASKS

• –––––––––––––––––––––––––––––––––––

• CLIENT

  • editing host/beacon/server name calls graphHandler.updateNodeName()
  • patterned bg for lifecycle rather than only grayscale?
  • 'timeless' / 'all time' mode to not show any time states

• LAYOUT

  • layout bugs
    • GroupNodes on top of each other, provide more room
    • space out subNodes in a group node more, they are very tight at high zoom scale
  • test layout and interaction with users
  • dragging pins nodes
  • persist pinned layout
  • methods for auto-layout
  • method to always expand hosts

• TIME STATE

  • preview nodes when they change in the time state
  • Maybe: preview all active? or just expand superNodes when there are active nodes?

• LAYOUT: SERVER RENDERING

  • with no group or sub nodes
    • host acts as node when interacted with...
    • possibly as its own ServerGraphRenderer
  • as a different shape?
  • rename isServer to isSingle or something agnostic

• CUSTOMIZE VISUAL

  • need UX mockups for this
    • should visuals be attached to comments or to graph entities?
  • icons for text labels
  • colors for nodes? outlines possibly? additional bonus circle within to style

• ALTERNATE LAYOUTS

  • time based simulation layout - for presentation mode
    • option to prioritize nodes based on presentation mode focus...
    • swarm plot
  • filter out nodes that
  • TRY: combine team servers into a single parent node?
    • give it special rendering and layout - possibly outside all other layout?

• UPDATE DATA

  • update with new dataset that has overlap with the old
    • test method to remove remove a host, beacon, whatever by ids
  • method to remove a host, beacon, whatever by id?

• LINK INTERACTION

  • preview and selection
  • labels
  • subdue all others while preview or selecting

• zoomToSelection - zoomToFit better

• graphData.SelectShortestPath()

• is dotGrid practical? what could an alternate, non-css approach be? possibly use d3 scales?

• improve Readme.md

• try a GIANT dataset

• childSelected class?

• rename time to lifeCycle?

• quadtree textOcclusion - if performance is bad

• FILTER NODES

  • Filter out nodes that are less connected
  • Or intensify nodes based on graph analytics like:
    • in degree
    • out degree
    • betweenness
    • centrality