>>main ideas similar to Dmitry and Naoto. Have seen Pack fail after hefty runtime, correlated with size of dbf, when earlier memory of the vfp process was taxed/fragmented.
>
>Perhaps a sys(1104) would help in that case?

Only to diagnose by roundabout reasoning, not to remedy. In that special case the memory became fragmented due to leakage originating in vfp itself, caused by some uuuhhhm, original coding practices, which nonetheless are not forbidden in vfp. Better way to diagnose is with some routines Calvin put in his blog years ago, allowing you to map the memory used by the application, showing fragmentation clearly.