The usage of boolean solids causes a segmentation fault when exiting an application and the problem has to do something with memory alocation. It seems like boolean solids overide other variables in memory. Sometimes some volumes dissapear from tracking (application /afs/cern.ch/user/t/tpapaeva/public/ProblemWithBooleanSolid.tar ), while the volumes exist in visualization (in this application volume called "Li-TargetFoil" and a "MylarFoil" is expected before "Li-TargetFoil". These volumes exist in visualization but not in tracking!). In another case (/afs/cern.ch/user/t/tpapaeva/public/ProblemWithBooleanSolid2.tar), the application seems to run properly but trying to visualize it causes core dump. In some other case the application was runing and terminating properly, but visualizing the geometry was causing "dissapearance" of some volumes in tracking and segmentation fault in exit. (Unfortunatly I dont have the code). In all cases the problem disapeard when replaced the boolean solids with something else.
The problem has been currently fixed in tag "geommng-V02-00-00" of geometry/management. The problem was in the destructor of G4SolidStore. Thanks for reporting this.
