Integer overflow w/ generated lexer

[netshade]

I was attempting to compile a lexer generated w/ the Rust backend, and I got the following errors on my tokens:

error: this arithmetic operation will overflow
     |
8379 | ...ze << (FLOAT - 1)) | (1usize << (VALID_IDENTIFIER - 1)) | (1usize << (FIELD - 1)) | (1usize << (INDEXED_FIELD - 1)))) != 0) {
     |                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ attempt to shift left by `61_isize`, which would overflow

I think this is because the template portion that is doing a bitset emulation of the Antlr code is presuming that 1usize is 64-bit width trying to match the original implementation?.

If my understanding here is correct, should it be that the template is instead masking with 1u64 instead ? Happy to provide a PR if this is right.

[netshade]

( This came about because I was compiling for wasm32-unknown-unknown - on other architectures it worked as expected )

[netshade]

( After looking at the template some more I see that this definitely is not quite so simple as that. Are the semantics of the parser more or less locked in at "everything must be word sized which is also 64 bits", or is it possible to be explicit that everything is 64 bits here? )