Map / Reduce engine

The engine in the runtime will be interfaced by the different SDKs in a declarative manner. Below is an example of how we see the C# version go.

Classify<GroupCreated>.AsAdd();
Classify<GroupDeleted>.AsRemove();

Classify<UserCreated>.AsAdd();
Classify<UserDeleted>.AsRemove();


Map
    .From<UserCreated>()
    .From<AddressSet>()
    .From<MobileNumberSet>()
    .From<UserDeleted>()
    .ReduceTo<TReadModel>(_ => {
        _.WithId<UserCreated>(e => e.UserId)
        _.WithPropertiesFrom<UserCreated>(readModel, source => {  // Should be possible to just specify eventType - no arguments, meaning it will map by convention (same name)
            readModel.Username = source.Username;
            readModel.Department = source.Department;
            readModel.UserNumber = _r.Count()    // We need operation classifications (Add / Remove) to be able to do this
        });
        _.WithPropertiesFrom<EmailSet>(readModel, source => {
            readModel.Email = source.Address;
        });
        _.WithPropertiesFrom<MobileNumberSet>(readModel, source => {
            readModel.MobileNumber = source.Number;
        });
        _.WithChildRelationShipDefinedBy<UserAddedToGroup>(_r => { // Work in progress.... 
                readModel.Groups = _.WithCollectionOf<UserGroup>(_c =>
                _c.WithpropertiesFrom<UserAddedToGroup>(readModel, source => {
                    readModel.GroupId = source.GroupId;
                }));
                _c.WithpropertiesFrom<GroupAdded>(readModel, source => {
                    readModel.Name = source.Name;
                }));
            })
        });
    })
    GroupBy(_ => _.Department, _g => {
        _g.ReduceToValue<UserCountPerDepartment>(_v => _v.Count())
    });