Graph Rendering Engines

[jmatsushita]

Having a discussion with @elf-pavlik and @Poietic about both RDF representation for metamaps and openoil and digressing into rendering engines and web components. Apparently https://famo.us/ is very performant (by using native APIs) on all devices. I've done a few performance tests and had also found ngraph. I was impressed by their pixi.js implementation (which was also very poor in capabilities, with no interaction and not even basic text rendering) but worked well to display thousands of nodes in a force directed physics simulation... I'll upload my test repo. I thought it could be for instance used for a "Navigation" panel with the whole graphed zoomed out, in a large graph.

We've also talked about existing UX patterns in commercial (or other) software and I'll create another issue for that!

Possibly of interest to @Connoropolous

[jmatsushita]

Also mentioned by @elf-pavlik https://github.com/noflo/noflo-ui uses web components (polymer specifically). It might not be the best example for manipulating large data sets, but maybe interesting for complex UI approaches.

I also did a bit of looking around with famo.us but haven't tried it yet, and it looks like there are interesting integrations including one with Meteor (what? there's also a meteor-polymer project? need to try all this!)

But otherwise it seems like it's own render tree paradigm (which looks a lot like a video compositing tree approach) might not map very well with a Web Components shadowDOM approach unless with some serious low level redesign... From my standpoint, unless I'm missing something, it's main power lies in using webGL and a motion graphics paradigm to building web/mobile experiences.

On the WebGL / Polymer front there's some Globe experiments with Three.js (with an ongoing attempt to make it into a Polymer component), force directed graphs in 3d

Also good to keep in mind sigmajs which uses WebGL like Famo.us does and is meant specifically for graphs. It also seems like it would be relatively simple to wrap sigma in a Polymer component...

Time for a table... (again)

[elf-pavlik]

ping @almereyda who also connected with Metacademy
example: http://metacademy.org/graphs/concepts/inference_in_mrfs#focus=gi2vjxnr&mode=explore

BTW wikimedia Berlin also hosts http://www.deepamehta.de/ meetups!