This is very early exploratory and will likely go nowhere.
-
My dissertation looked at Receiver Operating Characteristic... (mining my memory...) To get 100% True Positives identified, you have to accept 100% of false positives will be in the set. It got me thinking....
- Is there an analogy to ROC to get 100% compatibility on a site you have to accept 50% of APIs not useable because it's not in other browsers. Or to have 100% of people using your site, you will only have X% of web platform available to use..
- Can it describe the state of the ecosystem?
- (sorry for this framing) Gecko adds a feature before anyone else there's less incentive for developers to use it because the user-base is smaller. How would we help change this incentive?
- Chromium adds a feature before anyone else, there's a lot of usage of Chromium browsers - but ecosystem rightly wants to use things that are broadly compatible.
- WebKit adds a feature before anyone else US and EMEA tends to move because huge amounts of their revenue is generated via iOS + Safari.
- Actually do we know what developers say they support? Yes, Quarterly survey says: Chromium/Blink, WebKit, Gecko.
- Can it be used to help prioritize compat across all browsers? (More research needed)
- Can we look use caniuse compat data? WPT?
- Can we just look at wpt.fyi test runs?
- Why would Browser Vendors care? Work out what to work on. Show they are listening to the needs of developers.
- Why would Developers care? Perception, developers tell us constantly that compat is a huge issue in their day to day lives. There's a lot of work in browsers to improve that, but it goes unseen.
-
Can we use this to determine how safe it is to change an API, or remove it (not purely just usage counter).
Tools we can build:
- Compat Bot - look at wpt.fyi, see what tests have just now passed across the board and tweet success!
- Chromium Blog - automatically list compat improvements.