Changes from previous:
0.07012 2011-11-09 15:16:29
- as of 0.07011 all callbacks receive a ::Loader::Table or
interface-compatible object instead of the table name, this object
stringifies to the table name (RT#72260)
- fix a bug in dynamic schema_base_class/schema_components
implementation that ran the connection method twice on subsequent
connects
- use a temp file for filter_generated_code with a string program name
instead of IPC::Open2, which hangs on Win32 (RT#72226)
- previous version referred to the wrong RT# for the uniq_to_primary
change, it is actually (RT#51696)
0.07011 2011-11-01 09:00:00
- add -I option to dbicdump
- do not delete default custom content comment and ending 1; from custom
content in files that are being renamed (RT#70507)
- use MooseX::MarkAsMethods instead of namespace::autoclean for the
use_moose option, this protects operator overloads, only_autoclean
option added for the old behavior
- add experimental naming=v8 mode with better CamelCase identifier
support, relationship naming and conversion of non-identifier chars
(RT#71945)
- add naming => { force_ascii => 1 } option for Unicode database names
- implement schema_base_class and schema_components for dynamic and
working schemas
- remove dependency on File::Slurp
- allow the constraint and exclude options to be used simultaneously
(bphillips)
- fix Oracle multi-db_schema unique detection (RT#70851)
- fix Oracle common tests fail with multi_schema due to not resetting
the preserve_case option after the preserve_case tests (RT#70829)
- handle <type> DEFAULT NULL for Pg
- handle boolean DEFAULT 0::boolean for Pg
- config file support for dbicdump script (alnewkirk)
- added filter_generated_code option (RT#53841)
- generic table and column comments support
- MySQL table and column comments support
- support DOS line endings on *nix and *nix line ending on Win32
- add quiet option
- $schema->loader is now a public method
- add schema_components option
- sort relationships so they always come out in the same order
- also sort unique constraints so they always come out in the same order
- multi db_schema support with cross-schema rels (RT#39478)
- added moniker_parts option for name clashes in multi db_schema setups
- add rel_name_map option
- fix the decimal data type for MS Access over ODBC
- fix enum/set detection for MySQL (RT#68717)
- fix is_nullable detection on MS Access
- remove '$table has no primary key' warning
- added uniq_to_primary option to promote unique keys to primary keys
(RT#25944)
- support arrayrefs for result_namespace and resultset_namespace
(RT#40214)
- add naming => { monikers => 'preserve' } or 'singular'/'plural' to
control moniker inflection (RT#44935)
- add naming => { column_accessors => 'preserve' } to not normalize
CamelCase column names to lower case for accessors (RT#64668)
- support quoted PostgreSQL schema names with special chars (RT#64766)
- automatically turn on quoting for MySQL (RT#60469)
- become utf8-aware (RT#67920)
- handle duplicate relationship names (RT#64041)
- fix a bug in Sybase ASE foreign key detection
- generate POD for result_base_class, additional_classes,
additional_base_classes, left_base_classes, components,
result_components_map, result_roles, result_roles_map, unique
constraints, set_primary_key and table
- rename result_component_map to result_components_map (old name still
works)
- fix accessor collision detection for methods from
result_components_map components
- add result_roles and result_roles_map options
- fix for mysql rel detection in mixed-case tables on mixed-case
filesystems (OSX and Windows)
- support for DBD::Firebird
- support for unicode Firebird data types
- handle "use warnings FATAL => 'all';" in custom/external content
(RT#59849)
- for dynamic schemas, if the naming option is set, will automatically
turn on use_namespaces=1 as well. Set use_namespaces=0 to disable
this behavior (RT#59849)
0.07010 2011-03-04 08:26:31
- add result_component_map option
0.07009 2011-02-25 11:06:51
- fix a syntax error in MS Access ADO driver
0.07008 2011-02-25 01:54:43
- rename column_accessor_map to col_accessor_map, the old alias still
works
- support MSSQL over DBD::ADO
- support for MS Access over DBD::ODBC and DBD::ADO
0.07007 2011-02-15 10:00:07
- bump DBIx::Class dep to 0.08127
- fix MSSQL data types for native client and EasySoft driver
0.07006 2011-02-01 02:18:32
- turn unloading of RelBuilder temp classes back on, now with proper
check for class existance using Class::Inspector->loaded
- bump up dep on namespace::clean to avoid breakage with earlier
versions (RT#65149)
0.07005 2011-01-25 23:07:55
- support extra connect_info options like quote_char for dbicdump
- fix breakage on perl 5.8.x related to unloading temporary classes
0.07004 2011-01-24 03:43:05
- fix bug with result class methods being cached on in a closure instead
of the object, which breaks for multiple dynamic schemas in a single
perl instance
0.07003 2011-01-21 06:43:05
- fix relname/method collisions (RT#62648)
- fix fully qualified component classes (RT#62624)
- improve sybase/mssql db_schema detection
- remove MooseX::NonMoose from Schema files under use_moose=1
- better _tables_list for Sybase ASE
- add datetime_undef_if_invalid => 1 for MySQL datetime data types
(RT#64820) This behavior can be turned off by passing
datetime_undef_if_invalid=0 as a loader option
- added column_accessor_map option
- Preserve relationship names when redumping and another FK is added
(RT#62424)
- Remove resultset_components as ResultSetManager is deprecated
- Fix a fail when very old Moose/CMOP is installed
- Added warning for column-accessor collisions, doc section in ::Base
("COLUMN ACCESSOR COLLISIONS") and the col_collision_map option.
- Handle column accessor collisions with UNIVERSAL methods
- Generate custom_type_name hint for PostgreSQL enums, as used
by very recent SQL::Translator
- Added support for PostgreSQL enum types
- Added table/column comment support for Oracle
- Fix missing require (RT#62072)
pkgsrc changes:
- adjust dependencies
Upstream changes:
0.07002 2010-09-11 01:48:00
- Properly detect a schema loaded with use_moose on subsequent
reloads
- Die with a sensible message when a schema loaded with
use_moose => 1 is reloaded with use_moose => 0
- Switch to MRO::Compat
- Fix oracle common tests failure / lc(undef) warnings
- Bump Moose/Moosex::NonMoose optional dependencies to fixed-up
versions
- Fix mssql common tests failures with MSSQL 2005 (skip test of
datatypes found only on MSSQL 2008)
- Fix DB2 v8 test failures (skip tests of graphics types found
only on DB2 v9)
- Fix dangerous invocation of ->meta on classes during upgrade
(may be *non* moosified and contain a user-defined meta() )
- Multiple test cleanups and refactorings
to trigger/signal a rebuild for the transition 5.10.1 -> 5.12.1.
The list of packages is computed by finding all packages which end
up having either of PERL5_USE_PACKLIST, BUILDLINK_API_DEPENDS.perl,
or PERL5_PACKLIST defined in their make setup (tested via
"make show-vars VARNAMES=..."), minus the packages updated after
the perl package update.
sno@ was right after all, obache@ kindly asked and he@ led the
way. Thanks!
pkgsrc changes:
- adjust dependencies according to META.yml
Upstream changes:
0.07001 2010-07-24 21:28:08
- put is_deferrable => 1 back into default attributes for belongs_to
- fix Postgres sequence detection for qualified sequences
- detect DOS line ends in table/column comments and convert to \n
- added use_moose option
- always mark pk columns is_nullable=0
- fix unique constraint names for SQLite (actual names break ->deploy)
- fix bug in qualify_objects that would add schema to relnames
- better type info for Informix, except for DATETIME precision and
INTERVAL support
- better type info for DB2
- fix some newly-introduced test bugs
- fix composite PKs getting marked is_auto_increment on SQLite
0.07000 2010-05-22 23:40:15
- added qualify_objects option to prepend db_schema to table names
- fix for negative numeric default values
- sequence is detected for Oracle
- fix for SQLite is_auto_increment detection when table is empty (hobbs)
- rescan now reloads all tables
- minor type info improvements for all DBs
- fix erroneous default_value for MySQL NOT NULL columns (RT#57225)
- remove is_deferrable => 1 from default for belongs_to rels
- better type info for Oracle
- preliminary Informix support
- unregister dropped sources on rescan
- added 'preserve_case' option with support for all DBs where it makes
sense; removed the MSSQL 'case_sensitive_collation' and the
Firebird/InterBase 'unquoted_ddl' options in favor of it.
- support CamelCase table names and column names (in case-preserving
mode) at the v7 naming level
- rewrite datetime default functions as \'current_timestamp' where
possible (except for Sybase ASE) to ease cross-deployment
- use column_info instead of select to get Oracle column list (RT#42281)
- match quotes in MySQL parser in more places (RT#42101)
- fix unique detection in DB2 for multiple schemas (RT#39622)
- fix column name collisions with methods (RT#49443)
- fix loading MySQL views on older MySQL versions (RT#47399)
0.06001 2010-04-10 01:31:12
- fix type info for MSSQL
- fix MSSQL collation detection on freetds tds version 8.0
0.06000 2010-04-06 01:12:25
- better type info for MySQL
- initial MySQL data type tests (jhannah)
- don't set result_namespace if it's 'Result'
- support for MSSQL databases with case sensitive collation, manually
overridable with 'case_sensitive_collation' option
- do not try to detect driver and rebless when used with a custom
'loader_class'
- suppress 'bad table or view' warnings for filtered tables/views
- croak if several tables reduce to an identical moniker (ribasushi)
- better type info for Sybase ASE
- better type info for Pg: sets sequence for serials, handles numerics
without precision
- better _tables_list for MSSQL
- pick up views in SQLite too
- better rel inflection using Lingua::EN::Inflect::Phrase
- cascade_delete and cascade_copy are turned off for has_many/might_have
by default, and belongs_to has on_delete => 'CASCADE', on_update =>
'CASCADE' and is_deferrable => 1 by default, overridable via
relationship_attrs
- added config_file option for loading loader options from a file
- set inflate_datetime => 1 for 'AS getdate()' computed columns in
Sybase
- Firebird support
- use introspection pragmas instead of regexes to introspect SQLite
(hobbs)
- generate POD for refs correctly from column_info
- fix tables list, fk introspection and type info for SQL Anywhere
pkgsrc changes:
- Adjust dependencies
Upstream changes:
0.05003 2010-02-20 05:19:51
- support for custom_column_info, datetime_timezone and datetime_locale
(rbo)
- improve parsing of SQLite tables when a column definition
spans multiple lines (hobbs)
- fix missing trailing _id stripping for some relationship
names (rbuels)
- fixed accessor POD bug, was not dereferencing scalar refs
before printing (rbuels)
pkgsrc changes:
- Remove '*.orig' relics remaining from patch
Upstream changes:
0.05002 2010-02-15 10:17:47
- support for SQLAnywhere via DBD::SQLAnywhere and ODBC
- fix picking up quoted tables for SQLite (RT#54538) patch from schwern
- validate class/component loader_options to make sure classes
are available before generating the schema, patch from bphillips
0.05001 2010-02-05 14:29:27
- correct default_value for all backends with common tests
- fix bug with quoted Pg tables from $dbh->tables (RT#54338)
- add inflate_datetime => 0 to 'timestamp' types for Sybase
pkgsrc changes:
- Adjust license
- Add options for additional features
- Adjust dependencies
Upstream changes:
0.05000 2010-02-01 09:24:24
- better data_type, default_value and size for Sybase
- added 'generate_pod' option, defaults to on
- added 'pod_comment_mode' and 'pod_comment_spillover_length' to
control table comment generation (waawaamilk)
0.04999_14 2010-01-14 06:47:07
- use_namespaces now default, with upgrade/downgrade support
- filter out un-selectable tables/views
- fix NUMERIC/DECIMAL size column_info for postgres
- now mentions skip_load_external feature in comments (jhannah)
- moniker_map POD correction (jhannah)
0.04999_13 2010-01-03 12:32:25
- exclude 'size' column_info for postgres when unnecessary, and
use the correct precision for varying types (except NUMERIC)
- 'naming' attribute and backward compatibility with 0.04006
- added relationship_attrs option for setting attributes in
generated relationships
- added overwrite_modifications option that ignores md5sums on
generated code
- added skip_load_external (jhannah)
- remove Class::Data::Accessor and Class::Accessor::Fast and switch
everything to Class::Accessor::Grouped (jhannah)
- better handling of db_schema for Oracle, based on (RT#35732)
0.04999_12 2009-11-30 23:36:14
- fix MySQL rel introspection with on_connect_call =>
'set_strict_mode' (RT#52087)
- now using base 'DBIx::Class::Core' for Results (RT#52141)
0.04999_11 2009-11-29 18:08:46
- added patch to generate POD from postgres by Andrey Kostenko (GUGU)
- added test for norewrite feature
- fix default_value for MSSQL
0.04999_10 2009-10-31 12:28:53
- patch from Robert Bohne to make _table_uniq_info more correct for
Oracle
- fix data_type for identity columns with MSSQL
0.04999_09 2009-10-08
- Only redump the files when something has actually changed
- Place a warning at the top of the files saying 'do not modify' to
match the one at the bottom of the auto-gen'd section
0.04999_08 2009-08-28
- Replace UNIVERSAL::require with Class::C3::Componentised
- Add Sybase/MSSQL support through DBD::Sybase
- use $dbh->get_info(29/41) for qote_car/name_sep if available (semifor)
- add MSSQL support through DBD::ODBC
- support MSSQL table names with a '.' in the name
- support MySQL CURRENT_TIMESTAMP()
0.04999_07 2009-04-18
- Add result_base_class and schema_base_class options (RT #43977)
- Ignore duplicate uniq indices (including duplicates of the PK).
- Fix for DBD::SQLite 1.20
- Fix for DBIx::Class 0.08100
0.04999_06 Tue Nov 11, 2008
- Singularise table monikers by default
- Strip trailing _id from single-column belongs_to relationships
- Add "dbicdump" script for easy commandline dumping
- Throw out the in-memory class generation, just dump to a temporary
directory if the user didn't specify one
- Fix Oracle constraint and auto-increment detection for non-owned schemas
(RT #35732)
- Handle ResultSetManager deprecation warning in common tests
0.04999_05 Mon Apr 14, 2008
- Fix limiting table list to the specified schema for DB2
- Default db_schema to the username for DB2
- Allow specifying a custom loader_class, overriding the
storage_type-based detection
- Cosmetic fixes to dumping of externally defined classes
- Make ResultSetManager notice externally defined :ResultSet methods
- Fix test failure for non-InnoDB MySQL due to wrong skip count
- Fix base class ordering in dumped classes
- Run the common tests against both dynamic and dumped versions of
the schema
0.04999_04 Wed Mar 12, 2008
- Add is_auto_increment detecton for DB2
0.04999_03 Wed Mar 12, 2008
- Fix DB2 support
0.04999_02 Tue Feb 12, 2008
- Add is_auto_increment detection for Oracle
- Unnhide the Oracle module now that the CPAN perms are sorted
out. Thanks to Tsunoda Kazuya for the quick response.
0.04999_01 Tue Feb 5, 2008
- Mark foreign key columns with is_foreign_key => 1
- Add support for vendor-specific extra column attributes.
- Add support for extra => { unsigned => 1 } for MySQL.
- Add support for enum value lists for MySQL
- Set join_type => 'LEFT OUTER' for nullable foreign keys
(patch from Bernhard Wei??huhn)
- Set is_auto_increment for auto-increment columns (RT #31473)
(Only SQLite, MySQL and PostgreSQL are currently supported)
- Generate one-to-one accessors for unique foreign keys (ilmari)
- Add support for load_namespaces-style class layout
- Fix test skip count for main skip_rels block
- Fix auto-inc column creation for the Oracle tests
- Fix column ordering in unique constraints for Oracle
- Fix Win32 test skip counts for good (RT #30568, Kenichi Ishigaki)
- Default Oracle db_schema to db username (patch
from Johannes Plunien)
- Updating package for p5 module DBIx::Class::Schema::Loader from
0.04005 to 0.04006
- Setting license to gnu-gpl-v2
- Adjusting and reordering dependencies
Upstream changes:
0.04006 2009-05-01
- Fix Oracle constraint detection for non-owned schemas (RT #35732)
- Add result_base_class and schema_base_class options (RT #43977)
- Handle ResultSetManager deprecation warning in common tests
- Fix for DBIx::Class 0.081
to trigger/signal a rebuild for the transition 5.8.8 -> 5.10.0.
The list of packages is computed by finding all packages which end
up having either of PERL5_USE_PACKLIST, BUILDLINK_API_DEPENDS.perl,
or PERL5_PACKLIST defined in their make setup (tested via
"make show-vars VARNAMES=...").
DBIx::Class::Schema::Loader automates the definition of a
DBIx::Class::Schema by scanning table schemas and setting up columns and
primary keys.
DBIx::Class::Schema::Loader supports MySQL, Postgres, SQLite and DB2.
See DBIx::Class::Schema::Loader::Generic for more, and
DBIx::Class::Schema::Loader::Writing for notes on writing your own
db-specific subclass for an unsupported db.
This module requires DBIx::Class 0.05 or later, and obsoletes
DBIx::Class::Loader for DBIx::Class version 0.05 and later.
While on the whole, the bare table definitions are fairly straightforward,
relationship creation is somewhat heuristic, especially in the choosing
of relationship types, join types, and relationship names. The relationships
generated by this module will probably never be as well-defined as
hand-generated ones. Because of this, over time a complex project will
probably wish to migrate off of L<DBIx::Class::Schema::Loader>.
It is designed more to get you up and running quickly against an existing
database, or to be effective for simple situations, rather than to be what
you use in the long term for a complex database/project.
