See the strange text issue in #120.
If someone knows where the encoding needs to be specified and can test that, please shoot us a PR.

Seems like the following line of code is getting a carriage return while transforming legacy lang files to JSON:

It should instead be matching up to \n or \r\n.

@haykam821 Thanks! No idea why I didn't think of LangResourcePack. That'll probably be good enough as a workaround (or fix if you will).

Ideally building on both systems would produce jars with identical content though (disregarding timestamps and pseudo-random things). So I'll keep this issue open for that.