Groups experiment

[captainbrosset]

See the experiment live here which is based on this list of groups.

Rationale:

The current web-features groups are not very consistent and don't cover all features. Some groups are about language features, others are about APIs. That's because we never decided what groups should be.

Web-features are made by putting ourselves in developers' shoes. Each feature correspond to something a web developer would want to achieve by using the platform.

This proposal is about pushing that concept further, to include groups too.
In this proposal, groups correspond to things that developer want to do, from the very high level "make a website" down to much finer grain tasks like "draw to a canvas".

Going through this exercise of categorizing features showed me a few things:

• Such a hierarchical view of the platform is amazing for discovery. If we can mix this with some filtering based on baseline status, this can be very useful for developers to discover previously impossible capabilities, or refactor JS-heavy code with now-native APIs.
• Adding new groups over time is very easy. As new feature get added, patterns emerge and new group just make sense.
• It makes it quite obvious when features are not based on actual developer use cases, because they don't fit anywhere.
  E.g, I'm having a real hard time putting ARIA attribute reflection in a group.
• It makes it obvious when we're missing features too. For example I wanted to create a group about installing site as apps, but realized were didn't yet have any of the web app manifest features.

I'm opening this issue to invite collaboration on the tentative grouping I've made in featureCatalog.yml.

In my opinion, I think it is possible to come up with one way to group features based on developer use cases, as I've started here. But as discussed at the last WebDX meeting, I understand that there may be many different ways to group, depending on what you want to learn, and on your perspective. So I'm encouraging other opinions.