A partial archive of discourse.wicg.io as of Saturday February 24, 2024.

“Difficulty ratings” for proposals


So, when somebody proposes a spec, one of the most valuable pieces of feedback toward the creation of an action plan is a summary of what would be required to make the spec a reality. Some of the factors that feed into this:

  • Have there been existing discussions around this?
  • Are UAs already doing this, just in non-standard / non-exposed ways? (I was opposed to Shadow DOM until somebody explained that it was something all the major browsers already had infrastructure for.)
  • As a combination of the last two, is this already a sub-behavior of a larger specification that could just be made more versatile? Examples:
    • The deep clone algorithm, which is currently specified as part of specs like postMessage but not actually available as a standalone function
    • The way hiding whitespace-only text nodes is part of Flexbox
  • What would be involved in polyfilling it (usually this is a similar discussion to “Are UAs already doing it?”)

This is basically a Best Common Practices memo on the subject of what factors go into making a spec reality, though there could maybe be some kind of sub-group / tiger team within/alongside WICG (reaching out to UA devs) that goes around doing structured appraisals for this in discussions (ie. “As a heads up, this would be a Class C change because it’s not polyfillable and would require all-new logic to support”).