Best way to write fields of vectors in CZML?

[rsignell-usgs]

@cleder, as described https://github.com/TerriaJS/terriajs/issues/1063 we are trying to figure out how to get vector fields into TerriaJS, like this one: (See the full Jupyter notebook in case you are interested in how this was generated)

It has been suggested to use CZML, and I'm wondering whether it's best to take the approach suggested to use billboards (scaled/rotated icons), or whether we could use the agi_vector described here: https://github.com/AnalyticalGraphicsInc/cesium/wiki/CZML-Content#agi_vector

Do you have a suggestion of which approach to take? Could I use your czml python package to write either of these?

[Frencil]

@rsignell-usgs that's a really cool use case for Cesium! I can help address your immediate question.

My best guess is that this library will need some minor expansion to do what you need whether you use vectors or billboards.

Billboards

There is billboard support but the CZML specification for billboards has expanded somewhat since the last time this library was kept up-to-date with it. Issue #17 provides an example of where this library currently falls short. There are only a handful of newer attributes billboards can assume that this library lacks so this should be a relatively quick and straightforward expansion.

AGI_Vectors

Presently the library has no agi_vector support at all!

Adding support

While I'm sure you're looking for something that "just works" out of the box, as a somewhat recent collaborator on this project myself I would say that adding missing support for either missing billboard attributes or agi_vectors would not be terribly difficult. This library has very good test coverage and is structured well enough that figuring out how to make additions is relatively straightforward and actually making them just requires a bit of time with the CZML spec.

It's also worth emphasizing that a lot of the other plumbing to generate CZML documents, particularly the attributes that are used most commonly, already enjoy robust support with this library.

Unfortunately I can't promise to implement these missing features myself any time soon, so it would be up to you or another collaborator if you're looking for a faster turnaround. If you're interested in expanding support in either of these areas to fit your use case I would be happy to provide assistance and code review!

[rsignell-usgs]

@Frencil , thanks for the detailed feedback -- that definitely helped keep me from going down a rabbit hole on this. The "spending time with the CZML spec" is likely the time sink, I would guess. I'm more of an end-user oceanographer than a developer, unfortunately. I'll see what I can do though...