Support automatically generating Unicode property lists

[bakkot]

PRs like tc39/ecma262#1939 and tc39/ecma262#1896 could (and should) be done automatically as part of the tooling.

@mathiasbynens maintains a collection of npm modules which could be used for this. Note that these bump Unicode versions in minor releases, so for this use case we'd want to pin a minor version.

[bakkot]

Alternatively we could commit the relevant data files to ecma262 and add an option in ecmarkup to read a data file passed in on the command line. That probably makes more sense, so that we can update the version in the spec without having to mess with ecmarkup.

[ljharb]

i like passing it in; but I’d hope we don’t have to commit data to 262 to do it - couldn’t we pass the path to one of the npm packages you referenced?

[bakkot]

I'm not really a fan of dynamically loading dependencies, but yes, we probably could.

[mathiasbynens]

I agree that not committing data to tc39/ecma262 seems best.

I noticed this was filed as an ecmarkup issue. We could solve it as this level, or we could add a script to the package.json that outputs the *.html files containing the tables (e.g. npm run build-unicode-properties). The latter seems simpler.

Regardless of where we decide to solve this, we could use the following packages in particular to generate the tables:

• https://github.com/mathiasbynens/unicode-canonical-property-names-ecmascript
• https://github.com/mathiasbynens/unicode-property-value-aliases-ecmascript
• https://github.com/mathiasbynens/unicode-property-aliases-ecmascript