I am interested in regexp implementation (high level overview). As far as I understand, it is not possible to come up with state-machine for regex containing "extensions" such as backreference (\1 or \2), it just doesn't fit into "regular languages". At least this is what I learnt from [0].

Does anyone know more details about new regex matcher? (I don't think I will be able to read the code).

[0]: http://swtch.com/~rsc/regexp/