MonetDBLite installation and usage example

monetdblite.R

# Obsolete, See https://www.monetdb.org/blog/monetdblite-r

See https://www.monetdb.org/blog/monetdblite-r
You can use any directory for the second parameter to dbConnect(), provided its writeable. Then you can also keep the data you loaded around for another time.

Package: MonetDBLite
Version: 1.0.0
Title: In-Process version of MonetDB for R

I still can't believe it yet. You made it in just few months.
Is this really MonetDB server, that embedden in R? Or R embedded in MonetDB server?
So, with MonetDBLite, I can include MonetDBLite library inside my own program?

@aris it is a MonetDB database, embedded in R. You talk to it via R function calls wrapped a DBI interface. We also had R as a UDF language in MonetDB for a year or so now. But yes, with MonetDBLite, you can embed MonetDB in other programs. We also have a low-level C library, see https://dev.monetdb.org/hg/MonetDB/file/embedded/tools/embedded/embedded.h for the interface.

Thanks for all the work on the package. I noticed an issue I wanted to get your thoughts on. I have written R code that combs through csv files I want combined into a database table. The csv's are individually read and than appended to a sqlite database using essentially:

tempsheet <- read.csv(ith csv)
dbWriteTable(con, name = "tablename", value = tempsheet, append = TRUE)

As the script runs, the amount of avaliable RAM stays pretty constant. When I change this code to write to a monetdblite database instead of sqlite, the amount of available ram steadily decreases as the script runs. Any ideas on why this might be happening or tips on ways around it?

@iantist please note that there is a difference between allocated virtual memory and the resident set. MonetDB uses memory-mapped files to store data, this is expected behavior. Are you experiencing any issues due to the raising memory footprint?