Hi Stefan, Riham PF_ROX has the new indexing code, that Lefteris has tried to describe in his recent paper. This code was only checked for correct operation from MIL in the code generated by the pf-rox java program of Riham. The correct operation of the new indexing code when activated automatically from mps was not tested yet. It may well be the source of trouble. i will pick this up once I have some time again (..). Peter Dear all, I checked all (modified) files in the PF_ROX branch by hand, and all but runtime/pathfinder.mx & runtime/pf_support.mx (this means in particular the compiler/mil/milprint_summer.c I fixed the other day) are OK, i.e., they contain all (intended) changes applied in the PF_ROX branch as well as all chnages that were made on the development trunk since the creation of the PF_ROX branch (and hence properly propagated). For runtime/pathfinder.mx & runtime/pf_support.mx, I am not yet sure, whether everything is OK. There have been numerous "interleaved" (and partly "redundant" or "conflicting") changes both genuine on the PF_ROX branch and propagated from the development trunk, and the (very tedious) analysis requires more time than I could spent today. More to come "soon", I hope ... Unfortunately, while only 50 (100) tests fail in the new stable branch (and in the development trunk), more than 500 tests seem to fail in the PF_ROX branch --- also this requires "some" more ananlysis ... Stefan