I'm planning to create a (probably written in Vulkan, so it will be ambitious...) project where I want to leverage modern 3D graphics for displaying potentially very large graphs in an intuitive & interactive way. I'm trying to target use cases such as note taking and mind-mapping as well as code browsing and git repo browsing as I see many potential ways to create visualization capability (via unification) across the multiple abstraction layers (many of which involve graph structures). E.g., git tree is a DAG, and the nodes on there are source file trees themselves. We could view a list of GitHub PRs on a repo as a list of nodes which point into locations within the repo's git tree. It's turtles all the way down, since source files can themselves be decomposed into ASTs, and those yet further into syntactical trees.
I imagine that with common workflows (one of which I just proposed) without the barrier of context switching could be a big win. At a certain zoom level I'm envisioning being able to directly see which portions of source are touched by the different PRs for example.
Another example of this is in applying this toward file browsing and the management of backups (i'll want to sprinkle tree differencing visualization in there).
Some thoughts:
Source code should be able to fall back to editable text when necessary.
Multiple Projections to view/interface the same data with different views must be possible.
Optimal would be that it integrates any editor I'd like (I want sublime, others not so much). Ideally other file formats like word documents, latex documents etc. should also be supported. But that sounds very hard, basically makes the whole thing an infinte pan zoom graph based window manager (with all the other graph related features).
Hm.. Maybe this all really is on the operating system level.
I favor html as base technology for the reason of easily including different "apps", like code editors, pdf viewers, webgl content, etc.
But thats just because I don't really want to develop this at the window manager/operating system level... Honestly I feel html is too slow for what I really want. There is a reason I use sublime and not VSCode, even it is just a millisecond or so.
I hope more people think in similar directions with the graph based interface. May one day some hero actually succeed with what we envision.
I encourage you to be that one!
One more thing:
I am also toying with the idea of using Jupyter notebooks as a starting point. One could relatively easy derive relationships between code cells and display them in graph based view instead of linear sequence. Then go further and add projections for AST.
I imagine that with common workflows (one of which I just proposed) without the barrier of context switching could be a big win. At a certain zoom level I'm envisioning being able to directly see which portions of source are touched by the different PRs for example.
Another example of this is in applying this toward file browsing and the management of backups (i'll want to sprinkle tree differencing visualization in there).