RIOTDB: transparently bringing scalability and I/O-efficiency to statistical computing with R http://www.cs.duke.edu/dbgroup/Main/RIOT NOTE: This is only a prototype implementation for demonstrating research ideas. -------------------------- To install RIOTDB on a Unix-like system, follow these instructions: 1) Database setup. Currently RIOTDB uses MySQL as its backend. Please make sure you have access to a MySQL server. You can modify the server address, username, and password in the RIOTDB/src/clib/config.h file to match your settings. The default username RIOTDB assumes is 'mysql' and the default password is empty. Use the script RIOTDB/src/scripts/init_riotdb.sql to initialize and populate the database: $ mysql -u mysql # log in the mysql command interface $ \. <path_to_the_script> # load the script If any error should occur, make sure the user account has enough privileges. 2) Compile the package. Make sure you are at the top level directory (where this file resides), and type the following: $ R CMD INSTALL ROITDB The RIOTDB package will be installed. If you want to have it installed in a customized directory, make a file ~/.Renviron with the following contents before installing the package: R_LIBS=<path> 3) (Optional) Patch R. If you want to use an optimized version of RIOTDB (see the discussion at http://www.cs.duke.edu/dbgroup/Main/RIOT-interface), use the file R-RIOTDB.patch to patch the official R source and rebuild R. Any reasonably modern R release should work. $ cd <top level of R's source tree> $ patch -p1 < <path to the R-RIOTDB.patch file> $ ./configure # build R as usual $ make && make install 4) Use RIOTDB. Start R and use the following statement to load RIOTDB: > library(RIOTDB)