Pooled ODBC for Erlang
Switch branches/tags
Nothing to show
Pull request Compare This branch is even with pannonia-technologies:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Pooled ODBC for Erlang

podbc is a simple wrapper built on top of poolboy and the odbc library bundled with Erlang/OTP to pool ODBC connections in Erlang. In addition to the thin wrapper functions around the functions found in Erlang's odbc module, there are functions to reduce boilerplate code and make your life easier, when you use ODBC with Erlang. The project was formerly known as poolboy_odbc. The rename was made mainly to reduce typing.

Contributions are always very welcome! See here who contributed in the past.

I try to keep the API as stable as possible, but things can still change, before we reach version 1. So please be aware of that.



1> podbc:start().

Adding a pool programmatically:

2> podbc_mgr:add_pool('odbc/test', [{size, 5}, {max_overflow, 5}],
        [{dsn, "DSN=test"}, {options, [{auto_commit, off}, {binary_strings, on}]}]).

Removing a pool programmatically:

3> podbc_mgr:remove_pool('odbc/test').

Sample configuration file:

You can supply the configuration file with the -config option of erl. The example below will automatically configure the pool 'odbc/test' and start the podbc application.

$ erl -pa ebin -pa deps/*/ebin -config sample.config -s podbc start
%% sample.config
  {podbc, [
    {pools, [
        [{size, 5}, {max_overflow, 5}],
        [{dsn, "DSN=test"}, {options, [{auto_commit, off}, {binary_strings, on}]}]
    {mappings, [
      {'odbc/test', mysql}
    {named_queries, [
      {mysql, [
        {last_insert_id, "SELECT CONVERT(LAST_INSERT_ID(), UNSIGNED)", []},
        {'person.all', "SELECT id, last_name, first_name FROM person", []},
        {'person.insert', "INSERT INTO person (last_name, first_name) VALUES (?, ?)", [
            {sql_varchar, 64}, {sql_varchar, 64}]}

All methods of Erlang's odbc module are supported. See odbc for details. The functions connect and disconnect work a little different.


4> {ok, WorkerRef} = podbc:connect('odbc/test').


The API for disconneting changed. You only need to pass the WorkerRef, no need for two arguments. disconnect/2 is still there for backwards compability, but will be removed sometime in the future.

5> podbc:disconnect(WorkerRef).

do a way to reduce boilerplate and safely connect and disconnect:

The code below will get a connection from the pool, then pass the worker reference to the fun, execute the fun and then safely return the connection to the pool.

Inside the fun, we use the named_query/3 function. See the example configuration file above for the sql.

6> podbc:do('odbc/test', fun(WorkerRef) ->
        {selected, _, Result} = podbc:named_query(WorkerRef, 'person.all', []), Result end).

Example with inserting data:

7> podbc:do('odbc/nestor', fun(Worker) ->
        {updated, 1} = podbc:named_query(Worker, 'person.insert', [<<"John">>, <<"Doe">>]),
        {selected, _, [{Id}]} = podbc:named_query(Worker, last_insert_id, []),

Road Map:

  • Complete podbc_qresolver, to be able to reload query mappings.
  • Complete documentation.
  • Write tests! At the moment there are no tests at all, which is really, really bad...
  • Provide mock test library. So code using podbc can be easily tested.
  • Add an asynchronus interface to named_query, param_query and sql_query.
  • Add some kind of DDL generator, to automatically update the database schema on changes.



Here I just want to say thank you and give credit to people, who were contributing to the project in the past.

All contributors in chronological order of their first contribution.


podbc is licensed under the MIT license. See LICENSE for details.