Abstract away data sources with layouts #7

Components like timeline, glossary, issues and even dynamic-view have similar structures. They could be abstracted away to a single parametrizable component.

Note that screenstack itself follow that pattern even if it's not text. The layout itself can be done, as explored before, directly with e.g D3 or Cytoscape, or pre-computed, e.g Gephi.

It, being able to explore and improve the PIM from a wiki, is actually the main motivation of this project.

Consider different visualizations, e.g :

web-url, most demanding but most up to date and (partly) interactable (outside of VR)
screencast, less demanding (rendered server side) but static and also partial (no iframe, no video, etc)
"just" page name, very light but also very limited

Depending on different criteria zoom-level dependant visualization (as touched on in https://observablehq.com/@utopiah/d3-pim-graph#cell-1183 ), or ZUI, could be interesting to explore.

Consider a set of shortcuts, e.g on wrist, per layout type.

For example while trying to interact (to define, could be highlight when within a certain distance) with content of a graph shortcuts could be related to e.g edge.target() and edge.source() (cf https://js.cytoscape.org/#collection/traversing )

Keeping in mind already defined MIME types and Content-Type which might be too low-level. Also the recently encountered the DataTransferItem interface https://twitter.com/utopiah/status/1569929593028546560

See also https://fabien.benetou.fr/Analysis/LibrarianMoveWalls on the specifics of mapping data structure and explorable and memorable place with scaling constraints.

This also means that a set of jxr snippets would be displayed alongside the item of a type, e.g for a graph each node would have jxr snippets on highlight to toggle the visibility of connecteds nodes, for an edge graph one of such snippets pinching on it would teleport to the connect node.

Overall jxr snippets per type should be stored on dedicated page to be composable.

See also WebXR, even AFrame dataviz components, e.g BabiaXR https://gitlab.com/babiaxr/aframe-babia-components

Consider how gltf-jxr https://git.benetou.fr/utopiah/text-code-xr-engine/src/branch/gltf-jxr namely jxr code direclty in a glTF model could do so, i.e change its shape or behavior based on a live data source.

A trivial way to test this would be to react on a cyclic event a heart 3D model which change scale (i.e beats) every second, or something more specific applied within a model itself, a clock where the hands change their orientation based on the actual time of the user loading it.