RE: https://mastodon.social/@wingo/116294319311502266

Very interesting. A concern is that there is an implicit assumption that Test262 actually has 100% test coverage of the relevant Ecma specs. In theory, it should cover every externally observable requirement that is expressed in the spec algorithms and prose. That's was an asperational which as far as I know is yet to be achieved.

This suggests another interesting experiment: could Claude Code, starting with the specs, generate a new version of Test262 with more comprehensive coverage.