Fix ambiguous grammar

[wwerner]

We have some ambiguous rules for numbers, i.e. BYTE_LITERAL and DECIMAL_LITERAL, so antlr matches the one that was defined first.

In int answer = 42 , it matches the number as BYTE_LITERAL and complains that it requires an INT_LITERAL . Once the number contains underscores or exceeds the byte range, it matches correctly.

Now I'm wondering whether I missed sth. (antlr version/config) or if it is just an error. That was implemented before my time so I can't say whether there is a reason for it I don't understand or if it's just a bug.

[VaughnVernon]

@wwerner In Java a byte literal is an int that is cast to a byte. Thus, I think you should eliminate BYTE_LITERAL and use INT_LITERAL for both bytes and ints. The one difference is that you must check for min/max byte range, Byte.MIN_VALUE and Byte.MAX_VALUE.

[wwerner]

Some pointers from @VaughnVernon copied in from slack for future reference:

• https://github.com/antlr/grammars-v4
• https://github.com/antlr/grammars-v4/blob/f40d77f85ee8a299d309cdcd15bfff5041bc44ad/java/java/JavaParser.g4#L286
• https://github.com/antlr/grammars-v4/blob/f40d77f85ee8a299d309cdcd15bfff5041bc44ad/java/java/JavaLexer.g4#L87
• https://github.com/antlr/grammars-v4/blob/f40d77f85ee8a299d309cdcd15bfff5041bc44ad/java/java/JavaParser.g4#L372
• https://github.com/antlr/grammars-v4/blob/f40d77f85ee8a299d309cdcd15bfff5041bc44ad/java/java/JavaParser.g4#L584
• https://github.com/antlr/grammars-v4/blob/f40d77f85ee8a299d309cdcd15bfff5041bc44ad/java/java/JavaParser.g4#L588