lower casing rules necessary?

[BobHanson]

[creating an issue for Commit 1393c02]

This is an interesting point, and well taken that Locales can mess with String.toLowerCase(). As mentioned in the JavaDoc for String.toLowerCase(). It's not something I had thought about much. I'm sure I'm not alone. But, yes, this is an issue for the three Locale default languages tr (Turkish), az (Azerbaijani), and lt (Lithuanian).

It's probably a generally advisable recommendation for ALL Java programs to issue

Locale.setDefault(Locale.ROOT)

unless language-dependence is needed.

However, in this case it is unnecessary. We do not care what the lowerCase translates to, as we are just using the case to compare with a similarly lower-cased map key.

Or am I missing something here?

[dan2097]

It's probably a generally advisable recommendation for ALL Java programs to issue

Locale.setDefault(Locale.ROOT)

A Java program could get away with doing this, but a Java library mustn't do this as the program including the library may have locale dependent behaviour. It's an unfortunate language decision that various Java APIs don't default to locale-independent behaviour, but it is at least easy enough to override on a method by method basis cf. a similar issue in C IUPAC-InChI/InChI#100

However, in this case it is unnecessary. We do not care what the lowerCase translates to, as we are just using the case to compare with a similarly lower-cased map key.

In the Turkish locale the lower case of i is i as expected, but the the lower case of 'I' is 'ı' e.g. the following will return false

Locale turkishLocale = new Locale("tr", "TR");
Locale.setDefault(turkishLocale);
System.out.println("OneFiveT".toLowerCase().equals("ONEFIVET".toLowerCase()));

hence a lookup on ONEFIVET would fail to find the OneFiveT enum in this locale.