With OML4R datastores, you can save R objects in the database.

R objects, including OML4R proxy objects, exist for the duration of the current R session unless you explicitly save them. The standard R functions for saving and restoring R objects, save and load, serialize objects in R memory to store them in a file and deserialize them to restore them in memory. However, for OML4R proxy objects, those functions do not save the database objects associated with the proxy objects in an Oracle database; therefore the saved proxy objects do not behave properly in a different R session.

You can save OML4R proxy objects, as well as any R object, with the ore.save function. The ore.save function specifies an OML4R datastore. A datastore persists in the database when you end the R session. The datastore maintains the referential integrity of the objects it contains. Using the ore.load function, you can restore in another R session the objects in the datastore.

Using a datastore, you can do the following:

• Save OML4R and other R objects that you create in one R session and restore them in another R session.

• Pass arguments to R functions for use in Embedded R Execution.

• Pass objects for use in Embedded R Execution. You could, for example, use a function in the OREdm package to build an Oracle Machine Learning for SQL model and save it in a datastore. You could then use that model to score data in the database through Embedded R Execution. For an example of using a datastore in an Embedded R Execution function, see Example 9-6.

The following table lists the functions that manipulate datastores and provides brief descriptions of them.