Composition of Search Strategy with different limit

[bourreauEric]

Expected behavior

limit the first search strategy to 1 second in solve THEN use the second strategy (to improve the best so far solution)

Actual behavior

search.limit Monitor are attach to search and not strategy.

Minimal Working Example

Experienced with choco-solver-4.0.4

solv.setSearch( new AbstractStrategy[] { Search.intVarSearch(new MaxRegret(), new IntDomainMax(), X), Search.domOverWDegSearch(TableVariableCouleur)} ); solv.limitTime(1000);

can be

solv.setSearch( new AbstractStrategy[] { Search.intVarSearch(new MaxRegret(), new IntDomainMax(), X).limitTime(1000), Search.domOverWDegSearch(TableVariableCouleur).limitTime(10000} );

[jgFages]

good idea, can you do it?

[fagossa]

Hello, I would like to give this a try.

Is that ok?

[cprudhom]

@fagossa that's perfectly ok

[Xabibax]

If it's still open I would like to work on it. Just wich time limit should have priority in a case like this:

solv.setSearch(
new AbstractStrategy[] {
Search.intVarSearch(new MaxRegret(), new IntDomainMax(), X).limitTime(1000),
Search.domOverWDegSearch(TableVariableCouleur).limitTime(10000}
);
solv.limitTime(5000);

[cprudhom]

I think that, for design reason, the global time limit should always have the priority. Each strategy will only be concerned by its own limit (time, nodes, failures, ...).

And thank you for proposing yourself on the PR.