Need String#normalize

[paulmillr]

No description provided.

[Yaffle]

String.prototype.normalize = function(){return String(this);}

[ljharb]

Unfortunately it's far more complex than that. The spec says that it takes a "form" argument that defaults to "NFC", and it throws a RangeError if String(form) is not "NFC", "NFD", "NFKC", or "NFKD" - then it uses the Unicode Normalization Algorithm for the specified form.

[mathiasbynens]

http://git.io/unorm

[cscott]

The Unicode databases required are rather large; perhaps this should be an optional dependency? That is, something like:

String.prototype.normalize = function(form) { return require('unorm')[String(form).toLowerCase()](this); }

with the appropriate error handling and helpful messages if require('unorm') fails.

[ljharb]

https://github.com/walling/unorm/blob/master/lib/unorm.js doesn't seem that large… While that's a valid concern, I think having a single es6-shim.js file that shims as much as it can is very important to the purpose of the shim.

As for a require, that'd be nice, but we'd need to ensure the shim still works with all the module systems it currently has - it might be better to copy/paste it in, like I did with the Math.fround/numberConversion stuff.

[cscott]

https://github.com/walling/unorm/blob/master/lib/unorm.js is 141kB. That's the standalone version of the library (it appears to combine unorm.js with https://github.com/walling/unorm/blob/master/src/unormdata.js, which is 122kB).

For comparison, unminimized es6-shim.js is currently 42kB. So this would be a 435% increase in the size of es6-shim -- probably a larger increase if you look at the minimized sizes. I think that's large enough that people would avoid using es6-shim for mobile projects.

[ljharb]

Fair point. If we want to point people at https://github.com/walling/unorm, then we need to make it very clear in the documentation.

[paulmillr]

yup totally agree, including ultra-big shims with the lib is not good

at least until we will have custom builds

[cscott]

I added a pointer to unorm in the documentation (see c185084 and a04221b).

[ljharb]

Closing this for now - when/if #136 happens, we will include String#normalize as an includable option.