Objective

I sketch a build system aimed at sharing derived objects. The local dependency graphs are stable, and the global structure emerges from the composition.

I want to focus on build optimization issues, avoiding multiple checking of components often depended upon.

The techniques used tend to make build sessions behave as nested transactions, and could be moved to within clearmake. The concern is however to make it possible to use ClearCase as a tool for sharing contributions, and communicating about them.

Sharing is non-trivial and requires collaboration. Attention to detail makes it possible.

Trying to find the reasons, which explain that there is a demand for UCM:


RUC 2002 Paper ToC
Marc Girod
Last modified: Thu Jan 17 00:02:16 EET 2002