sqlQuery: Query an ODBC Database

On success, a data frame (possibly with 0 rows) or character string. On error, if errors = TRUE a character vector of error message(s), otherwise an invisible integer error code -1

(general, call odbcGetErrMsg for details) or -2

(no data, which may not be an error as some SQL statements do return no data).

sqlQuery is the workhorse function of RODBC. It sends the SQL statement query to the server, using connection channel returned by odbcConnect, and retrieves (some or all of) the results via sqlGetResults.

The term ‘query’ includes any valid SQL statement including table creation, alteration, updates etc as well as SELECTs. The sqlQuery command is a convenience wrapper that first calls odbcQuery and then sqlGetResults. If finer-grained control is needed, for example over the number of rows fetched, additional arguments can be passed to sqlQuery or the underlying functions called directly.

sqlGetResults is a mid-level function. It is called after a call to sqlQuery or odbcQuery to retrieve waiting results into a data frame. Its main use is with max set to non-zero when it will retrieve the result set in batches with repeated calls. This is useful for very large result sets which can be subjected to intermediate processing.

Where possible sqlGetResults transfers data in binary form: this happens for columns of (ODBC) SQL types double, real, integer and smallint, and for binary SQL types (which are transferred as lists of raw vectors, given class "ODBC_binary"). All other SQL data types are converted to character strings by the ODBC interface.

This paragraph applies only to SQL data types which are returned by ODBC as character vectors. If when creating the connection (see odbcConnect) DBMSencoding was set to a non-empty value, the character strings are re-encoded. Then if as.is is true for a column, it is returned as a character vector. Otherwise (where detected) date, datetime and timestamp values are converted to the "Date" or "POSIXct" class. (Some drivers seem to confuse times with dates, so times may get converted too. Also, some DBMSs (e.g. Oracle's) idea of date is a date-time.) Remaining cases are converted by R using type.convert. When character data are to be converted to numeric data, the setting of options("dec") is used to map the character used by the ODBC driver in setting decimal points---this is set to a locale-specific value when RODBC is initialized if it is not already set.

Using buffsize will yield a marginal increase in speed if set to no less than the maximum number of rows when believeNRows = FALSE. (If set too small it can result in unnecessarily high memory use as the buffers will need to be expanded.)

Modern drivers should work (and work faster, especially if communicating with a remote machine) with rows_at_time = 100, the usual default, or more. (However, some drivers may mis-fetch multiple rows, in which case set rows_at_time = 1 when creating the connection.) However, if max is specified then this may fetch too many rows and hence it could be reduced (but then this setting applies to all subsequent fetches from that result set). Another circumstance in which you might want to reduce rows_at_time is if there are large character columns in the result set: with the default value up to 6Mb of buffer for each such column could be allocated to store intermediate results.