next up previous contents
Next: Subprocess descriptors Up: DISK/DIRECTORY SYSTEM Previous: Disk space control


No facility was provided for entering ECS system objects into directories. In fact, the underlying ECS system was destroyed daily and recreated the next day. Thus it was impossible to conceive of a directory containing ECS objects. As a way around this difficulty, two types of disk objects were provided, static and dynamic name-tags. Name-tags could be associated with ECS system objects, and actions were provided to obtain a capability for the object associated with a given name-tag. Static name-tags always referred to the same (or isomorphic) ECS system objects; and dynamic name-tags could be associated with arbitrary ECS objects by user programs. (In fact, the association was with capabilities, hence not restricted to ECS objects.) Elaborate conventions were required by the user programs to make sure that the dynamic name-tags were in fact associated with appropriate ECS objects, since the association was destroyed daily. The system used name-tags to provide directory entries for the ECS files and event channels associated with I-O devices. During each system startup, the appropriate associations were reconstructed.

Paul McJones