A-Frame is pioneering, making architectural choices which may or may not pan out. Think Angular 1.
Neither WebVR nor UnityVR are currently usable on Linux. Think "I got my simple test scene to not core dump!". MacOS has almost nothing. Android has some support from both, and iOS less, but I'm unfamiliar with them. There's a lot of Windows-centrism in the current VR community - if someone says something "works", they likely mean on Windows, with android a distant second. Once upon a time, Microsoft would write MS-only non-standard web extensions... now Google and Mozilla are doing the writing for them.
So, I generally recommend: Unity on Windows for VR beginners and exploration; Unity or https://www.unrealengine.com/ (better Linux support) for more serious work. But there are other options. And for WebVR, Windows (or Android?), A-Frame for beginners, direct three.js for more serious work, and perhaps aframe-react for exploratory VR webdev.
Aside: And for a hypothetical never-met person who wishes to run a Vive with WebVR on Linux, there's also my https://github.com/mncharity/node-webvr-alt-stack ... but I didn't finish WebVR 1.1 and A-Frame support.
Surprisingly capable in terms of what you can do -- if you're willing to get your hands dirty and debug your way out THREE.js corners. Nonetheless surprising considering it's all JS running in the browser, and it definitely works.
Unsurprisingly immature in terms of ecosystem and support you should expect. Unsurprising considering WebVR isn't even a standard (yet), and browser support isn't mainline.
If you want to ship an app, it's not in the same ballpark as Unity at this point. But if you want to explore what Unity might look like if it was written in JS in the browser, give it a try.
This coming from someone who vehemently does believe.
Much younger obviously, but same entity-component architecture and dedicated to WebVR. A lot of components just like there are on the Unity Asset Store.
