The SQL Server adapter for ActiveRecord using SQL Server 2012 or higher.
Interested in older versions? We follow a rational versioning policy that tracks Rails. That means that our 7.x version of the adapter is only for the latest 7.x version of Rails. If you need the adapter for SQL Server 2008 or 2005, you are still in the right spot. Just install the latest 3.2.x to 4.1.x version of the adapter that matches your Rails version. We also have stable branches for each major/minor release of ActiveRecord.
Adapter Version | Rails Version | Support |
---|---|---|
7.0.0.0 |
7.0.x |
active |
6.1.2.1 |
6.1.x |
active |
6.0.2 |
6.0.x |
active |
5.2.1 |
5.2.x |
ended |
5.1.6 |
5.1.x |
ended |
4.2.18 |
4.2.x |
ended |
4.1.8 |
4.1.x |
ended |
For older versions, please check their stable branches.
We support every data type supported by FreeTDS. All simplified Rails types in migrations will correspond to a matching SQL Server national (unicode) data type. Always check the initialize_native_database_types
(here) for an updated list.
The following types (date
, datetime2
, datetimeoffset
, time
) all require TDS version 7.3
with TinyTDS. We recommend using FreeTDS 1.0 or higher which default to using TDSVER
to 7.3
. The adapter also sets TinyTDS's tds_version
to this as well if non is specified.
The Rails v5 adapter supports ActiveRecord's datetime_with_precision
setting. This means that passing :precision
to a datetime column is supported. Using a precision with the :datetime
type will signal the adapter to use the datetime2
type under the hood.
The adapter uses OUTPUT INSERTED
so that we can select any data type key, for example UUID tables. However, this poses a problem with tables that use triggers. The solution requires that we use a more complex insert statement which uses a temporary table to select the inserted identity. To use this format you must declare your table exempt from the simple output inserted style with the table name into a concurrent hash. Optionally, you can set the data type of the table's primary key to return.
adapter = ActiveRecord::ConnectionAdapters::SQLServerAdapter
# Will assume `bigint` as the id key temp table type.
adapter.exclude_output_inserted_table_names['my_table_name'] = true
# Explicitly set the data type for the temporary key table.
adapter.exclude_output_inserted_table_names['my_uuid_table_name'] = 'uniqueidentifier'
Although it is not necessary, the Ruby convention is to use lowercase method names. If your database schema is in upper or mixed case, we can force all table and column names during the schema reflection process to be lowercase. Add this to your config/initializers file for the adapter.
ActiveRecord::ConnectionAdapters::SQLServerAdapter.lowercase_schema_reflection = true
Depending on your user and schema setup, it may be needed to use a table name prefix of dbo.
. So something like this in your initializer file for ActiveRecord or the adapter.
ActiveRecord::Base.table_name_prefix = 'dbo.'
It's also possible to create/change/drop a schema in the migration file as in the example below:
class CreateFooSchema < ActiveRecord::Migration[7.0]
def up
create_schema('foo')
# Or you could move a table to a different schema
change_table_schema('foo', 'dbo.admin')
end
def down
drop_schema('foo')
end
end
We currently conform to an unpublished and non-standard AbstractAdapter interface to configure connections made to the database. To do so, just override the configure_connection
method in an initializer like so. In this case below we are setting the TEXTSIZE
to 64 megabytes.
module ActiveRecord
module ConnectionAdapters
class SQLServerAdapter < AbstractAdapter
def configure_connection
raw_connection_do "SET TEXTSIZE #{64.megabytes}"
end
end
end
end
TinyTDS supports an application name when it logs into SQL Server. This can be used to identify the connection in SQL Server's activity monitor. By default it will use the appname
from your database.yml file or your Rails::Application name.
Below shows how you might use the database.yml file to use the process ID in your application name.
development:
adapter: sqlserver
appname: <%= "myapp_#{Process.pid}" %>
Every class that sub classes ActiveRecord::Base will now have an execute_procedure class method to use. This method takes the name of the stored procedure which can be a string or symbol and any number of variables to pass to the procedure. Arguments will automatically be quoted per the connection's standards as normal. For example:
Account.execute_procedure(:update_totals, 'admin', nil, true)
# Or with named parameters.
Account.execute_procedure(:update_totals, named: 'params')
The 3.2 version of the adapter support ActiveRecord's explain features. In SQL Server, this is called the showplan. By default we use the SHOWPLAN_ALL
option and format it using a simple table printer. So the following ruby would log the plan table below it.
Car.where(id: 1).explain
EXPLAIN for: SELECT [cars].* FROM [cars] WHERE [cars].[id] = 1
+----------------------------------------------------+--------+--------+--------+----------------------+----------------------+----------------------------------------------------+----------------------------------------------------+--------------+---------------------+----------------------+------------+---------------------+----------------------------------------------------+----------+----------+----------+--------------------+
| StmtText | StmtId | NodeId | Parent | PhysicalOp | LogicalOp | Argument | DefinedValues | EstimateRows | EstimateIO | EstimateCPU | AvgRowSize | TotalSubtreeCost | OutputList | Warnings | Type | Parallel | EstimateExecutions |
+----------------------------------------------------+--------+--------+--------+----------------------+----------------------+----------------------------------------------------+----------------------------------------------------+--------------+---------------------+----------------------+------------+---------------------+----------------------------------------------------+----------+----------+----------+--------------------+
| SELECT [cars].* FROM [cars] WHERE [cars].[id] = 1 | 1 | 1 | 0 | NULL | NULL | 2 | NULL | 1.0 | NULL | NULL | NULL | 0.00328309996984899 | NULL | NULL | SELECT | false | NULL |
| |--Clustered Index Seek(OBJECT:([activerecord... | 1 | 2 | 1 | Clustered Index Seek | Clustered Index Seek | OBJECT:([activerecord_unittest].[dbo].[cars].[P... | [activerecord_unittest].[dbo].[cars].[id], [act... | 1.0 | 0.00312500004656613 | 0.000158099996042438 | 278 | 0.00328309996984899 | [activerecord_unittest].[dbo].[cars].[id], [act... | NULL | PLAN_ROW | false | 1.0 |
+----------------------------------------------------+--------+--------+--------+----------------------+----------------------+----------------------------------------------------+----------------------------------------------------+--------------+---------------------+----------------------+------------+---------------------+----------------------------------------------------+----------+----------+----------+--------------------+
You can configure a few options to your needs. First is the max column width for the logged table. The default value is 50 characters. You can change it like so.
ActiveRecord::ConnectionAdapters::SQLServer::Showplan::PrinterTable.max_column_width = 500
Another configuration is the showplan option. Some might find the XML format more useful. If you have Nokogiri installed, we will format the XML string. I will gladly accept pathches that make the XML printer more useful!
ActiveRecord::ConnectionAdapters::SQLServerAdapter.showplan_option = 'SHOWPLAN_XML'
NOTE: The method we utilize to make SHOWPLANs work is very brittle to complex SQL. There is no getting around this as we have to deconstruct an already prepared statement for the sp_executesql method. If you find that explain breaks your app, simple disable it. Do not open a github issue unless you have a patch. Please consult the Rails guides for more info.
When creating a new Rails application you can specify that you want to use the SQL Server adapter using the database
option:
rails new my_app --database=sqlserver
To then connect the application to your SQL Server instance edit the config/database.yml
file with the username, password and host of your SQL Server instance.
The adapter has no strict gem dependencies outside of ActiveRecord
. You will have to pick a connection mode, the default is dblib which uses the TinyTDS
gem. Just bundle the gem and the adapter will use it.
gem 'activerecord-sqlserver-adapter'
If you would like to contribute a feature or bugfix, thanks! To make sure your fix/feature has a high chance of being added, please read the following guidelines. First, ask on the Gitter, or post a ticket on github issues. Second, make sure there are tests! We will not accept any patch that is not tested. Please read the RUNNING_UNIT_TESTS
file for the details of how to run the unit tests.
- Github: http://github.com/rails-sqlserver/activerecord-sqlserver-adapter
- Gitter: https://gitter.im/rails-sqlserver/activerecord-sqlserver-adapter
Many many people have contributed. If you do not see your name here and it should be let us know. Also, many thanks go out to those that have pledged financial contributions.
You can see an up-to-date list of contributors here: http://github.com/rails-sqlserver/activerecord-sqlserver-adapter/contributors
Copyright © 2008-2022. It is free software, and may be redistributed under the terms specified in the MIT-LICENSE file.