With this change the solver during install now considers only the part
of the tree which is required to calculate the solution, it doesn't
consider anymore World() as the search space.
The search space now is narrowed down to the packages that related to
the one which we are considering.
In this subset of changes we are also optimizing the Parallel solver
avoiding an useless loop.
This change boost overall performance on large datasets which don't
necessarly have relations touching the whole tree.
This happens because we select the best instead of selecting the best match in Package BuildFormula()
- Extend the selection test
- Select from databases when ordering
- Relax assertions search
- Split compile step in test
- Adapt tests
Note: This is temporarly until we treat this case in BuildFormula() inside Package
we need to build the constraints between all the requires when expanding
and then create a new constraint that selects the best match
Fixes races conditions and make the DB Switchable. Also prepare inside
the CompilationSpec the tree of the deps to be built, and parallelize
only the building jobs.
Closes#7
Signed-off-by: Ettore Di Giacinto <mudler@gentoo.org>
Split dependency resolving ( definition <-> db mapping ) into a method of the tree (ResolveDeps),
and do not care of looking up for packages while parsing.
Note: it is slower but it is much accurate and we can
choose what to do with the ones which aren't resolved (and show a helpful message too).
