In my use case I'm generating data at runtime, and the cost of serializing such data into a CSV file and then loading to MonetDBLite using COPY will be potentially greater than using something like INSERT INTO tablename VALUES. Given that MonetDBLite does not use DBConnections as in ODBC/iODBC, does that mean that I do not need to clear the memory using explicit COMMIT statements, or do I still need to run COMMIT after loading some GB of data into a table in order not to occupy all the available primary memory? Thanks in advance.
Which MonetDBLite are you using? You can use from-memory appends for very fast data loading in this case I think. For R/Python, we can directly append from data frames, for C the process is a little less convenient.