project

project copied to clipboard

https://github.com/mdn/browser-compat-data/issues/15083 is a proposal to simplify maintenance of BCD by drastically reducing the amount of manually maintained data, in the typical case inferring the data for 5 browsers (Chrome Android, Edge, Opera, Opera Android, WebView Android) from the Chrome data.

This project would consistent of multiple parts:

• [x] The schema and code changes to support mirroring (#109)
• [x] Replacing manual data with mirrored data where they would match (#110)
• [ ] Vetting places where manual data remains and is likely to be wrong

There would be no change to the published packages, which makes it all simpler / less risky.

@queengooborg can you add the "proposed for Q3 2022" label for this?

I actually don't have editor access for this repository, so I can't add labels!

@teoli2003 how about you?

I have, so I added the label 14 hours ago :-)

Oops, thanks and sorry for the noise!

It's my understanding that Samsung is not a perfect mirror of Chrome. They fork from a version earlier than stable and sometimes merge features from later versions. I would suggest confirming that all of the Chromium browsers can be mirrored.

@jpmedley It's not, and that is all the more reason why this project would be helpful. By mirroring the majority of the Chromium data over to the derivative browsers, such as Samsung Internet, the focus can be placed on the differences between the browsers, meaning less time must be spent on BCD's maintenance!

Phase one of this project has now been implemented in https://github.com/mdn/browser-compat-data/pull/16401! There are some bugs to work out before we perform a mass conversion of existing data to use mirroring.

I've taken off the "proposed for Q3'22" label and added this as a Q2'22 project because we're already working on it now.

What remains of the original checklist is "Vetting places where manual data remains and is likely to be wrong". @queengooborg do you have thoughts on how we should go about this? What is "likely" to be wrong? Off-by-one differences in mirroring are the obvious case, but are the other things to vet?

I'd like to close this project as complete since BCD has mirroring now. Please open a new OWD project proposal if there is a follow up project from here.