mccs (1)

mccs [options]

mccs (which stands for Multi Criteria CUDF Solver) is a solver for package dependency problems expressed in the CUDF format. By default, mccs reads a problem specification from standard input, and writes the solution to standard output.

-i file

Read input from file file instead of standard input.

-o file

Write the solution to file instead of standard output.

-fo

full solution output

-v n

set verbosity level to n.

-h

print this help

By default, mccs uses the cbc solving engine.

-lpsolve

use lpsolve solver

-lp lpsolver

specify a solving engine that takes the cplex input format. lpsolver is the path of a script that takes as input the file name containing the cplex input, and that produces the solution on standard output (an example for scip is given in /usr/share/doc/mccs/engines/sciplp).

-pblib pbsolver

use pseudo-Boolean solver pbsolver as solving engine.

-nosolve

do not solve the problem (for debugging)

-noreduce

do not reduce the initial problem

-only-agregate-constraints

generate only agregate constraints

-only-desagregate-constraints

generate only deagregate constraints (default)

-all-constraints

generate all kind of constraints (ensure redundancy)

See the file /usr/share/doc/mccs/README.optimization-criteria for a full grammar of optimization criteria.

An example input file can be found at /usr/share/doc/mccs/examples/legacy.cudf.

    mccs -i legacy.cudf

calls mccs on examples/legacy.cudf and prints the solution on stdout. With such a call mccs will resort to the default underlying solver cbc and use a default criterion to solve the problem.

    mccs -i legacy.cudf -o sol -lexagregate[-removed,-changed]\
    -lpsolve

Here, mccs puts the solution into the file "sol" and solves the problem using the lpsolve solver with the paranoid criterion, which consists of first minimizing the number of removed packaged, and then the number of packages that change installation status or installed version.

    mccs -i legacy.cudf -o sol \
    -lexsemiagregate[-removed,-notuptodate,\
    -nunsat[recommends:,true],-new]

The criterion used here is the trendy criterion, which consists of first minimizing the number of package removals, then to minimize the number of installed packages that are installed in a version older the most recent available version, then to minimize the number of recommendations of installed packages that are not satisfied, and finally the number of newly installed packages.

mccs has been written by Claude Michel <[email protected]>. The development of mccs has been partly supported by the European research project Mancoosi.