When a job is run on the server, the engine working directory is set to a new, empty "transient" directory.

TERR or open-source R code can write data to an empty transient directory by specifying non-absolute file paths. For example, executing the expression cat(1:10, file="foo.txt") writes a new file foo.txt in this directory. Also, you can use this working directory for temporary files.

After job finishes, this "transient" directory is kept around for awhile, and then it is deleted. The time is controlled by the server configuration. The default setting for this property is 24 hours. (See your System Administrator for more information about changing this property.)

Files from the transient data directory can be read using WebDAV or plain HTTP. When a job finishes, the result includes a ResultDir containing the URL for accessing this directory. A client application can use this property to construct a complete URL for accessing a particular file in the directory.