Message10325

CPython and Jython both share the same implementation of the sre regex engine, which the re module uses by compiling to sre bytecodes.

Our current port of sre dates back to CPython 2.1, then subsequently updated. However, there are enough subtle differences - missing sre bytecodes, unimplemented optimzations for unicode, direct use of the Java stack, and for complex REs, divergent performance - that we should revisit.

Also, this will be a good opportunity to stop expanding unicode into codepoint arrays.