go-oracledb

go-oracledb is a package which is a database/sql/driver.Driver for connecting to Oracle DB, using Anthony Tuininga's excellent OCI wrapper, ODPI-C.

At least Go 1.11 is required!

Although an Oracle client is NOT required for compiling, it is at run time. One can download it from https://www.oracle.com/database/technologies/instant-client/downloads.html

Connect

In sql.Open("oracledb", connString), you can provide the classic "user/passw@service_name" as connString, or an URL like "oracle://user:passw@service_name".

You can provide all possible options with ConnectionParams. Watch out the ConnectionParams.String() does redact the password (for security, to avoid logging it - see go-goracle/goracle#79). So use ConnectionParams.StringWithPassword().

More advanced configurations can be set with a connection string such as: user/pass@(DESCRIPTION=(ADDRESS_LIST=(ADDRESS=(PROTOCOL=tcp)(HOST=hostname)(PORT=port)))(CONNECT_DATA=(SERVICE_NAME=sn)))

A configuration like this is how you would add functionality such as load balancing across mutliple servers. The portion described in parenthesis above can also be set in the SID field of ConnectionParams.

For other possible connection strings, see https://oracle.github.io/node-oracledb/doc/api.html#connectionstrings and https://www.oracle.com/pls/topic/lookup?ctx=dblatest&id=GUID-B0437826-43C1-49EC-A94D-B650B6A4A6EE .

TL;DR; the short form is username@[//]host[:port][/service_name][:server][/instance_name], the long form is (DESCRIPTION= (ADDRESS=(PROTOCOL=tcp)(HOST=host)(PORT=port)) (CONNECT_DATA= (SERVICE_NAME=service_name) (SERVER=server) (INSTANCE_NAME=instance_name))).

To use heterogeneous pools, set heterogeneousPool=1 and provide the username/password through oracledb.ContextWithUserPassw.

Rationale

With Go 1.9, driver-specific things are not needed, everything (I need) can be achieved with the standard database/sql library. Even calling stored procedures with OUT parameters, or sending/retrieving PL/SQL array types - just give a oracledb.PlSQLArrays Option within the parameters of Exec!

The array size of the returned PL/SQL arrays can be set with oracledb.ArraySize(2000)

• the default is 1024.

Connections are pooled by default (except AS SYSOPER or AS SYSDBA).

Speed

Correctness and simplicity is more important than speed, but the underlying ODPI-C library helps a lot with the lower levels, so the performance is not bad.

Queries are prefetched (256 rows by default, can be changed by adding a oracledb.FetchRowCount(1000) argument to the call of Query), but you can speed up INSERT/UPDATE/DELETE statements by providing all the subsequent parameters at once, by putting each param's subsequent elements in a separate slice:

Instead of

db.Exec("INSERT INTO table (a, b) VALUES (:1, :2)", 1, "a")
db.Exec("INSERT INTO table (a, b) VALUES (:1, :2)", 2, "b")

do

db.Exec("INSERT INTO table (a, b) VALUES (:1, :2)", []int{1, 2}, []string{"a", "b"})

Logging

Go-oracledb uses github.com/go-kit/kit/log's concept of a Log function. Either set oracledb.Log to a logging function globally, or (better) set the logger in the Context of ExecContext or QueryContext:

db.QueryContext(oracledb.ContextWithLog(ctx, logger.Log), qry)

Tracing

To set ClientIdentifier, ClientInfo, Module, Action and DbOp on the session, to be seen in the Database by the Admin, set oracledb.TraceTag on the Context:

db.QueryContext(oracledb.ContextWithTraceTag(oracledb.TraceTag{
	Module: "processing",
	Action: "first",
}), qry)

Extras

To use the go-goracledb-specific functions, you'll need a *oracledb.conn. That's what oracledb.DriverConn is for! See z_qrcn_test.go for using that to reach NewSubscription.

Calling stored procedures

Use ExecContext and mark each OUT parameter with sql.Out.

Using cursors returned by stored procedures

Use ExecContext and an interface{} or a database/sql/driver.Rows as the sql.Out destination, then either use the driver.Rows interface, or transform it into a regular *sql.Rows with oracledb.WrapRows, or (since Go 1.12) just Scan into *sql.Rows.

For examples, see Anthony Tuininga's presentation about Go (page 39)!

Caveats

sql.NullString

sql.NullString is not supported: Oracle DB does not differentiate between an empty string ("") and a NULL, so an

sql.NullString{String:"", Valid:true} == sql.NullString{String:"", Valid:false}

and this would be more confusing than not supporting sql.NullString at all.

Just use plain old string !

NUMBER

NUMBERs are transferred as oracledb.Number (which is a string) to Go under the hood. This ensures that we don't lose any precision (Oracle's NUMBER has 38 decimal digits), and sql.Scan will hide this and Scan into your int64, float64 or string, as you wish.

For PLS_INTEGER and BINARY_INTEGER (PL/SQL data types) you can use int32.

CLOB, BLOB

From 2.9.0, LOBs are returned as string/[]byte by default (before it needed the ClobAsString() option). Now it's reversed, and the default is string, to get a Lob reader, give the LobAsReader() option.

If you return Lob as a reader, watch out with sql.QueryRow, sql.QueryRowContext ! They close the statement right after you Scan from the returned *Row, the returned Lob will be invalid, producing getSize: ORA-00000: DPI-1002: invalid dpiLob handle.

So, use a separate Stmt or sql.QueryContext.

For writing a LOB, the LOB locator returned from the database is valid only till the Stmt is valid! So Prepare the statement for the retrieval, then Exec, and only Close the stmt iff you've finished with your LOB! For example, see z_lob_test.go, TestLOBAppend.

TIMESTAMP

As I couldn't make TIMESTAMP arrays work, all time.Time is bind as DATE, so fractional seconds are lost. A workaround is converting to string:

time.Now().Format("2-Jan-06 3:04:05.000000 PM")

See #121.

Install

Just

go get github.com/go-goracle/go-oracledb

Or if you prefer dep

dep ensure -add github.com/go-goracle/go-oracledb

and you're ready to go!

Note that Windows may need some newer gcc (mingw-w64 with gcc 7.2.0).

Contribute

Just as with other Go projects, you don't want to change the import paths, but you can hack on the library in place, just set up different remotes:

cd $GOPATH.src/github.com/go-goracle/go-oracledb
git remote add upstream https://github.com/go-goracle/go-oraceldb.git
git fetch upstream
git checkout -b master upstream/master

git checkout -f master
git pull upstream master
git remote add fork git@github.com:mygithubacc/go-oraceldb
git checkout -b newfeature upstream/master

Change, experiment as you wish, then

git commit -m 'my great changes' *.go
git push fork newfeature

and you're ready to send a GitHub Pull Request from github.com/mygithubacc/go-oraceldb, newfeature branch.

pre-commit

Add this to .git/hooks/pre-commit (after go get github.com/golangci/golangci-lint/cmd/golangci-lint)

#!/bin/sh
set -e

output="$(gofmt -l "$@")"

if [ -n "$output" ]; then
	echo >&2 "Go files must be formatted with gofmt. Please run:"
	for f in $output; do
		echo >&2 "  gofmt -w $PWD/$f"
	done
	exit 1
fi

golangci-lint run

Third-party

• oracall generates a server for calling stored procedures.