Hallo, I'm a bit puzzled by the failure of the compose_test test, which shows up in the current Metacomm results. I'm using BBv1 for running my tests, while I would guess that the Metacomm people are using BBv2. Still the generated command lines appear very similar, the only differences being the use of absolute paths and the explicit reference to the standard include directory in the Metacomm command line. I wonder if there's any configuration mismatch, like the use of different kinds of libraries (e.g. static vs. dynamic?). I'm checking to see if there are other diverging results. It would be interesting to know what other Borland users get from running the tests. Cheers, Nicola Musatti