For some users and use cases, it is tedious/impractical to create options objects when calling driver methods that take them. We should provide overloads taking bsoncxx::document::view_or_value objects to make the driver more versatile and usable.

Consider a case where the user has a bson document that they have stored options in, either generated from some config file or read out of the database:

// aggregate options
{ "maxTimeMS" : 100, "explain" : true, "allowDiskUse" : false }

In order to run an aggregation query, these options must first be converted to an options::aggregate object:

bsoncxx::document::value agg_options_bson{some existing bson};

options::aggregate opts{};
opts.max_time_ms(agg_options_bson["maxTimeMS"].get_int64());
opts.explain(agg_options_bson["explain"].get_bool());
opts.allow_disk_use(agg_options_bson["allowDiskUse"].get_bool());

collection.aggregate(pipeline, opts);

If we provided an overload we could skip the options-building steps here and just pass in the original document:

bsoncxx::document::value agg_options_bson{some existing bson};
collection.aggregate(pipeline, agg_options_bson);

Edit:

Other methods that we should consider changing, in addition to collection::aggregate():

• collection::create_index()