Open cahuja opened 7 years ago
Yeap, I read it the same way. Let's file UTS bug.
Or is it an ICU bug?
Perhaps @srl295 could shed some light...
@rxaviers it's a great question, can you send a note to cldr-users? I can confirm as with the links below:
icu4j yMd@en-SG dd/MM/y
icu4j yyMd@en-SG d/M/yy
Ok thanks @srl295
Matching to close patterns has different behavior in ICU4j and Globalize and here is an example of the difference. Consider the pattern "yyMD" in locale "en-SG"
As per CLDR data, the following pattern exists -
Now, when this same pattern (yMd) is run through Globalize and ICU4j, we get the expected result which is "dd/MM/y"
However, when I pass the pattern (yyMd) though Globalize and ICU4j, the results differ - Globalize (yyMd) -> dd/MM/yy ICU4j (yyMd) -> d/M/yy
When I read TR-35, I see that the notion of distance between y and yy would apply in a similar manner to how it would apply for MMM and MMMM as given in the example. In this case, I believe Globalize should be correct, but I would like another opinion before being sure to mark this as a spec bug.
Thoughts @rxaviers?