[Spec Bug] Requested options should expand best match options in DateTime Format

[tc39]

[Spec Bug] Requested options should expand best match options in DateTime Format

[caridy]

/cc @NorbertLindenberg @ericf @rwaldron @rxaviers @zbraniecki @srl295, can you guys look at this problem? I might be missing something, but it seems to solve all the issues with the polyfill. This issue was brought to me via andyearnshaw/Intl.js#145 (comment)

[caridy]

Here is the diff in the polyfill:

@@ -1935,7 +1935,9 @@ function/* 12.1.1.1 */InitializeDateTimeFormat (dateTimeFormat, locales, options
                 p = bestFormat[prop];

             // ii. Set the [[<prop>]] internal property of dateTimeFormat to p.
-            internal['[['+prop+']]'] = p;
+            // Diverging from spec because of bug #58
+            // https://github.com/tc39/ecma402/issues/58
+            internal['[['+prop+']]'] = opt['[['+prop+']]'] || p;
         }

[rxaviers]

I'm ccing @JCEmmons, who has implemented this for ibm-js/ecma402 if I'm not wrong, and could have input.

[srl295]

@caridy going to have to come back to this… 

[littledan]

For what it's worth, V8 follows the ICU algorithm, rather than the spec, here. It seems like Firefox does as well.

[caridy]

Further investigation reveal that BasicFormatMatcher doesn't have enough information to enforce the extra rule described here, which is fine, since this is the bare minimum required for this implemention, while BestFitFormatMatcher() can certainly take that into consideration. Closing.