NAME
DBIx::Class::OptimisticLocking - Optimistic locking support for
DBIx::Class
VERSION
version 0.02
SYNOPSIS
This module allows the user to utilize optimistic locking when updating
a row.
Example usage:
package DB::Main::Orders;
use base qw/DBIx::Class/;
__PACKAGE__->load_components(qw/OptimisticLocking Core/);
__PACKAGE__->optimistic_locking_strategy('dirty'); # this is the default behavior
PURPOSE
Optimistic locking is an alternative to using exclusive locks when you
have the possibility of concurrent, conflicting updates in your
database. The basic principle is you allow any and all clients to issue
updates and rather than preemptively synchronizing all data
modifications (which is what happens with exclusive locks) you are
"optimistic" that updates won't interfere with one another and the
updates will only fail when they do in fact interfere with one another.
Consider the following scenario (in timeline order, not in the same
block of code):
my $order = $schema->resultset('Orders')->find(1);
# some other different, concurrent process loads the same object
my $other_order = $schema->resultset('Orders')->find(1);
$order->status('fraud review');
$other_order->status('processed');
$order->update; # this succeeds
$other_order->update; # this fails when using optimistic locking
Without locking (optimistic or exclusive ), the example order would have
two sequential updates issued with the second essentially erasing the
results of the first. With optimistic locking, the second update (on
$other_order) would fail.
This optimistic locking is typically done by adding additional
restrictions to the "WHERE" clause of the "UPDATE" statement. These
additional restrictions ensure the data is still in the expected state
before applying the update. This DBIx::Class::OptimisticLocking
component provides a few different strategies for providing this
functionality.
CONFIGURATION
optimistic_locking_strategy
This configuration controls the main functionality of this component.
The current recognized optimistic locking modes supported are:
* dirty
When issuing an update, the "WHERE" clause of the update will
include all of the original values of the columns that are being
updated. Any columns that are not being updated will be ignored.
* version
When issuing an update, the "WHERE" clause of the update will
include a check of the "version" column (or otherwise configured
column using optimistic_locking_version_column). The "version"
column will also be incremented on each update as well. The
exception is if all of the updated columns are in the
optimistic_locking_ignore_columns configuration.
* all
When issuing an update, the "WHERE" clause of the update will
include a check on each column in the object regardless of whether
they were updated or not.
* none (or any other value)
This turns off the functionality of this component. But why would
you load it if you don't need it? :-)
optimistic_locking_ignore_columns
Occassionally you may elect to ignore certain columns that are not
significant enough to detect colisions and cause the update to fail. For
instance, if you have a timestamp column, you may want to add that to
this list so that it is ignored when generating the "UPDATE" where
clause for the update.
optimistic_locking_version_column
If you are using 'version' as your optimistic_locking_strategy, you can
optionally specify a different name for the column used for version
tracking. If an alternate name is not passed, the component will look
for a column named "version" in your model.
EXTENDED METHODS
update
See DBIx::Class::Row::update for basic usage.
Before issuing the actual update, this component injects additional
criteria that will be used in the "WHERE" clause in the "UPDATE". The
criteria that is used depends on the CONFIGURATION defined in the model
class.
_track_storage_value
This is a method internal to DBIx::Class::Row that basically serves as a
predicate method that indicates whether or not the orginal value of the
row (as loaded from storage) should be recorded when it is updated.
Typically, only primary key values are persisted but for
DBIx::Class::OptimisticLocking, this list is augmented to include other
columns based on the optimistic locking strategy that is configured for
this DBIx::Class::ResultSource. For instance, if the chosen strategy is
'"dirty"' (the default), every column's original value will be tracked
in order to generate the appropriate "WHERE" clause in any subsequent
"UPDATE" operations.
_storage_ident_condition
This is an internal method to DBIx::Class::PK that generates the "WHERE"
clause for update and delete operations.
BUGS
Please report any bugs or feature requests to
"bug-dbix-class-optimisticlocking at rt.cpan.org", or through the web
interface at
<http://rt.cpan.org/NoAuth/ReportBug.html?Queue=DBIx-Class-OptimisticLoc
king>. I will be notified, and then you'll automatically be notified of
progress on your bug as I make changes.
SUPPORT
You can find documentation for this module with the perldoc command.
perldoc DBIx::Class::OptimisticLocking
ACKNOWLEDGEMENTS
Credit goes to the Java ORM package <Hibernate> for inspiring me to
write this for DBIx::Class.
AUTHOR
Brian Phillips <bphillips@cpan.org>
COPYRIGHT AND LICENSE
This software is copyright (c) 2011 by Brian Phillips.
This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.