I believe JITting (code generation) happens at a later stage, after the source code had been parsed (or perhaps even "tokenized"/"lex'ed" is more appropriate?)

I believe JITting (code generation) happens at a later stage, after the source code had been parsed (or perhaps even "tokenized"/"lex'ed" is more appropriate?) So even an "unreliable" syntax wouldn't matter much since the jitter kicks in after the parser had delivered its results.