While I have been using yum the way it was intended for about a month now, I am interested in exploring a possibility of using yum as a dependency resolution "component," as part of a larger system. I have a way to textually specify system installation. Part of it lists all the packages installed on the system. The idea is that an any given time the textual representation if valid. Now, suppose a use edits this representation and adds one more package to the list. If I am to take this new representation and install a machine based on it, I would have no guarantee that it will be stable, because the user might not have specified all the dependent packages. To prevent this problem, I would like to provide an interface to the user to modify this textual representation. When user tell the system that he wants to add one package, the system might end up adding more to the list (textual representation of the system configuration) after all the dependencies are resolved. Since yum already does dependency resolution I would like to use it. Ideally, once the user specified the package he wants to add, internally, I run "yum -C get_dependency_set <user_package>" and yum returns, say: " <user_package> <package1_that_user_package_also_needs> <some_other_required_package> " I would like to run yum from cache, to make it a little faster. The "get_dependency_set" is, obviously, not implemented yet. What is the best way to achieve this? If the community thinks that this is a worthwhile flag to have, I would write the glue code necessary to get this flag call yum's internal functions (once I figure out where these functions are). Once other way would be to run "yum install <package> and just parse yum's output, since it has package information when it asks confirmation questions. But his seems like a bad way. -Sergei