[gnulinux]

Makes the grammar simpler which makes new implementations to implement tokenizer much faster/cheaper. It's a simple problem with simple solution that has been causing a medium sized issue, so why not fix it?