MapBox recently announced that they would be moving away from CartoCSS towards a completely visual interface. Apparently the new style sheets that underlay that UI will be JSON based, so despite the fact that the new interface is a point and click system one of the developers that worked on this project believes that it will ultimately be better for coders than the previous system:
the new editor https://t.co/i7uy2V8YGL is even better for coders: the GL styles can be modified as JSON in editors, and dynamically in JS
— Tom MacWright (@tmcw) July 29, 2015
It's worth noting however (and hopefully I have this right) that these JSON stylesheets are rendered by WebGL, while CartoCSS is used to create XML that is rendered by Mapnik. So while one is a replacement for the other, at least in some capacity, they don't do exactly the same thing.
So this leads me to wonder where does that leave CartoCSS?
Many of the predominant map tile styles use the language at this time, perhaps most notably OpenStreetMap's default style. CartoDB uses it as well and as one of their developers noted in the tweet below they plan to stick with it for the time being, but it also seems to imply that they will move away from it at some point.
@Anonymaps @tmcw we however are sticking with it a lilttle longer!
— Francisco Dans (@fdans) July 29, 2015
On a more recent front there's a project that called kosmtik which I found through this thread. This tool reads the same .mss (cartocss) and .mml (json) files as TileMill and renders with Mapnik so it does provide an alternative to TileMill which is no longer being actively developed.
The ultimate source of this question is that I undertook a pretty complex project a month or two ago in TileMill, and am perhaps 25% to 50% towards completing it. But with this announcement among other things I'm beginning to wonder if I'm hitching myself to a medium that won't be maintained/developed going forward.
So my question is, what other evidence exists that speaks to level of use of CartoCSS (or lack thereof) going forward by the GIS community at large?