new features about Number.prototype.toLocaleString() for IE11, a lot of related new definitions

[Clayblockunova]

I've finally made all these local Infinity and NaN stuff building normally. Is there anything else to do for us?

[Clayblockunova]

when will this be merged? should I wait for cleaning up work finished?

[fasttime]

Hi! I'm in the process of preparing version 3. I'd suggest holding off on any code cleanup until that's done. Once the new version is ready, I'll take care of this PR.

[Clayblockunova]

how does this PR work with current build?

[Clayblockunova]

can we merge this now?

[fasttime]

I've merged the parts with RUSSIAN_INFINITY and IE_11_WIN_8. I'll review the other changed when I find some time.

[Clayblockunova]

seems remarks generation need a fix？

[fasttime]

seems remarks generation need a fix？

Yes, I will take a look.

[fasttime]

Okay, the remarks should be fixed now.

[Clayblockunova]

can this be combined now?

[fasttime]

Sorry, this is a big PR and it may take some time for me to review all the changes. I've added some more comments.

[Clayblockunova]

anything else to do?

[Clayblockunova]

seems generation of ‎Features.md also need fix?

[Clayblockunova]

seems generation of ‎Features.md also need fix?

can I fix that? or should I leave it to you?

[fasttime]

Thanks! The descriptions in Features.md should be fixed now. I also left some questions on the PR.

[fasttime]

Is there a reason why only the first letter is used in definitions?

[Clayblockunova]

cuz only definition of the first letter practical in IE11 (there is no ITERATOR_HELPER in IE 11).

[fasttime]

Actually, it's okay to add definitions for feature combinations that aren't available in supported engines. Engines keep adding and removing features all the time, so some of those definitions may actually become useful at some point.

[fasttime]

It seems that ń and ć could be also defined from this string.

[Clayblockunova]

again, impractical in IE11.

[fasttime]

Similarly for 'zh' and zh-cn, maybe more characters could be defined?

[Clayblockunova]

again, only definitions of "正" and "字" are practical. btw, "非" and "数" are overlapped with 'ja' case.

[fasttime]

Oh, true. I hadn't noticed that some characters already have a definition.

[Clayblockunova]

anyway, how can I test whether a definition work under arbitrary feature set (including ones impossible in major environment?)

[fasttime]

anyway, how can I test whether a definition work under arbitrary feature set (including ones impossible in major environment?)

You need to use emuEval from feature-emulation.helpers. It's not included in the npm package, so you have to copy it from the repo. For example:

const JScrewIt = require('jscrewit');
require('./feature-emulation.helpers.js');

const definition = '([].at.constructor + [])[10]';
const features = ['AT', 'IE_SRC'];

const { replacement } = JScrewIt.debug.createEncoder(features).resolve(definition);
const value = emuEval(features, replacement);
console.log(value); // "F"

This works also inside browsers if you use <script> tags instead of require.

[Clayblockunova]

and, how can I filter overperformed definition automatically? can node dev/char-index.mjs do that?

[fasttime]

and, how can I filter overperformed definition automatically? can node dev/char-index.mjs do that?

node dev/char-index.mjs list :overdefined: will only tell you if any characters have unnecessary definitions. You can check .jscrewit.char-map.json to see which definitions are used and which not. Note that the characters must be added first (with node dev/char-index.mjs add <chars>).

[Clayblockunova]

seems "zh" cases is considered as overperformed by node dev/char-index.mjs. should we modify the function useLocaleNumeralDefinition()?

[Clayblockunova]

can optimize: true be used in define._callWithFeatures()?

[Clayblockunova]

I tried to add optimize: true in useLocaleNumeralDefinition(), and that works.

[Clayblockunova]

can this be merged now?