Because the production version of the MIDL compiler is not yet complete, it is premature to draw conclusions about MIDL's success in regard to usefulness and efficiency. However, we very much expect that MIDL will be a suitable tool for developing a production version of a global SETL optimizer from the (already formidable) SETL version of this optimizer. More generally, it will make SETL more widely useable by allowing critical sections of SETL programs to be optimized manually. PL/I and ALGOL 68 are existing languages with roughly the same semantic level as MIDL; However, MIDL is compatible with our existing SETL software, and is highly transportable.