tagline: From openSUSE
Solving dependencies is the core functionality for any software management application on Linux.
Dependencies are used to divide and share functionalities across multiple software elements.
Linux (and Unix) follows two basic concepts to implement Divide & Conquer - breaking a large problem into smaller, more manageable ones.
- Make each program do one thing well. See Basics of the Unix Philosophy for details.
- Use shared libraries to share implementations and save disk and memory space
Every software package expresses the functionalities it provides to others and those it requires from others through dependencies.
The task of the dependency resolver is to check and fulfill all these relations when managing software.
Basic laws of resolving
The dependency solver tries to solve dependencies without user intervention based on two basic rules
- Fulfill the install/remove requests given at start
- Keep the (dependencies of the) installed system consistent
Since the solver treats every package alike, these rules have some major and sometimes unexpected implications. A broken dependency might result in removal of lots of packages - the resulting system is still consistent by highly unusable.
Background & Information
The sat solver implementation as it appears in openSUSE 11.0 is based on two major, but independent, blocks
- Using a dictionary approach to store and retrieve package and dependency information.
- Using satisfiability, a well known and researched topic, for computing package dependencies.
Tools and libraries using SAT solver.