r - data.tables: merge

I have a second question on data.tables. As far as I have understood, merges are called joins in data tables. How can I control which type of merge I have (one-to-one, many-to-one, one-to-many), and whether the variables in the 'using' dataset will replace the variables in the master dataset?

Also, if keys are necessary in order to perform the merge, and I have to do more than one merge on my data, do I have to keep changing the keys? This appears not very clean to me ....

Thanks you in advance,

Matteo