X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=Changes;h=c2bb298d18ebddae4bfb0585de5b053e8bf29341;hb=f064a2abb15858bb39a141ad50391d4191988d2c;hp=274345eed1108388e5923c435a47fe9e298e830d;hpb=9930caaf7e7c250d914cb1440d9a0f1dd2a1dedc;p=dbsrgits%2FDBIx-Class.git diff --git a/Changes b/Changes index 274345e..c2bb298 100644 --- a/Changes +++ b/Changes @@ -1,13 +1,392 @@ +Current Known Issues / Regressions + - Breaks DBIx::Class::ResultSet::WithMetaData (fix pending in RT#104602) + - Breaks DBIx::Class::Tree::NestedSet (fix pending in RT#114440) + Revision history for DBIx::Class + * Notable Changes and Deprecations + - The entire class hierarchy now explicitly sets the 'c3' mro, even + in cases where load_components was not used. Extensive testing led + the maintainer believe this is safe, but this is a very complex + area and reality may turn out to be different. If **ANYHTING** at + all seems out of place, please file a report at once + - The unique constraint info (including the primary key columns) is no + longer shared between related (class and schema-level) ResultSource + instances + - Neither exception_action() nor $SIG{__DIE__} handlers are invoked + on recoverable errors. This ensures that the retry logic is fully + insulated from changes in control flow, as the handlers are only + invoked when an error is leaving the DBIC internals to be handled by + the caller (n.b. https://github.com/PerlDancer/Dancer2/issues/1125) + (also fixes the previously rejected RT#63874) + - $result->related_resultset() no longer passes extra arguments to + an underlying search_rs(), as by design these arguments would be + used only on the first call to ->related_resultset(), and ignored + afterwards. Instead an exception (detailing the fix) is thrown. + - Increased checking for the correctness of the is_nullable attribute + within the prefetch result parser may highlight previously unknown + mismatches between your codebase and data source + - Calling the set_* many-to-many helper with a list (instead of an + arrayref) now emits a deprecation warning + - Calling the getter $rsrc->from("argument") now throws an exception + instead of silently discarding the argument + + * New Features + - InflateColumn::DateTime now accepts the ecosystem-standard option + 'time_zone', in addition to the DBIC-only 'timezone' (GH#28) + - Massively optimised literal SQL snippet scanner - fixes all known + slowdowns ( in some cases 50x ) of very complex prefetch/selects + - DBIx::Class::Optional::Dependencies now properly understands + combinations of requirements and does the right thing with e.g. + ->req_list_for([qw( rdbms_oracle ic_dt )]) bringing in the Oracle + specific DateTime::Format dependencies + + * Fixes + - Fix regresion (0.082800) of certain calls being presented stale + result source metadata (RT#107462) + - Fix incorrect SQL generated with invalid {rows} on complex resultset + operations, generally more robust handling of rows/offset attrs + - Fix incorrect $storage state on unexpected RDBMS disconnects and + other failure events, preventing clean reconnection (RT#110429) + - Make sure exception objects stringifying to '' are properly handled + and warned about (GH#15) + - Fix incorrect data returned in a corner case of partial-select HRI + invocation (no known manifestations of this bug in the field, see + commit message for description of exact failure scenario) + - Fix corner case of stringify-only overloaded objects being used in + create()/populate() + - Fix spurious warning on MSSQL cursor invalidation retries (RT#102166) + - Fix several corner cases with Many2Many over custom relationships + + * Misc + - Add explicit test for pathological example of asymmetric IC::DT setup + working with copy() in t/icdt/engine_specific/sybase.t (GH#84) + - Fix t/54taint.t failing on local::lib's with upgraded Carp on 5.8.* + - Fix invalid variable names in ResultSource::View examples + - Skip tests in a way more intelligent and speedy manner when optional + dependencies are missing + - Make the Optional::Dependencies error messages cpanm-friendly + - Incompatibly change values (not keys) of the hash returned by + Optional::Dependencies::req_group_list (no known users in the wild) + - Stop using bare $] throughout - protects the codebase from issues + similar (but likely not limited to) P5#72210 + - Config::Any is no longer a core dep, but instead is migrated to a new + optdep group 'config_file_reader' + +0.082840 2016-06-20 07:02 (UTC) + * New Features + - When using non-scalars (e.g. arrays) as literal bind values it is no + longer necessary to explicitly specify a bindtype (this turned out + to be a mostly useless overprotection) + + * Fixes + - Ensure leaving an exception stack via Return::MultiLevel or something + similar produces a large warning + - Another relatively invasive set of ::FilterColumn changes, covering + potential data loss (RT#111567). Please run your regression tests! + - Ensure failing on_connect* / on_disconnect* are dealt with properly, + notably on_connect* failures now properly abort the entire connect + - Fix use of ::Schema::Versioned combined with a user-supplied + $dbh->{HandleError} (GH#101) + - Fix parsing of DSNs containing driver arguments (GH#99) + - Fix silencing of exceptions thrown by custom inflate_result() methods + - Fix complex prefetch when ordering over foreign boolean columns + ( Pg can't MAX(boolcol) despite being able to ORDER BY boolcol ) + - Fix infinite loop on ->svp_release("nonexistent_savepoint") (GH#97) + - Fix spurious ROLLBACK statements when a TxnScopeGuard fails a commit + of a transaction with deferred FK checks: a guard is now inactivated + immediately before the commit is attempted (RT#107159) + - Fix the Sybase ASE storage incorrectly attempting to retrieve an + autoinc value when inserting rows containing blobs (GH#82) + - Remove spurious exception warping in ::Replicated::execute_reliably + (RT#113339) + - Work around unreliable $sth->finish() on INSERT ... RETURNING within + DBD::Firebird on some compiler/driver combinations (RT#110979) + - Fix leaktest failures with upcoming version of Sub::Quote + - Really fix savepoint rollbacks on older DBD::SQLite (fix in 0.082800 + was not sufficient to cover up RT#67843) + + * Misc + - Test suite is now officially certified to work under very high random + parallelism: META x_parallel_test_certified set to true accordingly + - Typo fixes from downstream debian packagers (RT#112007) + +0.082821 2016-02-11 17:58 (UTC) + * Fixes + - Fix t/52leaks.t failures on compilerless systems (RT#104429) + - Fix t/storage/quote_names.t failures on systems with specified Oracle + test credentials while missing the optional Math::Base36 + - Fix test failures when DBICTEST_SYBASE_DSN is set (unnoticed change + in error message wording during 0.082800 and a bogus test) + - Remove largely obsolete test of SQLite view deployment (RT#111916) + + * Misc + - Work around rare test deadlock under heavy parallelism (RT#108390) + +0.082820 2015-03-20 20:35 (UTC) + * Fixes + - Protect destructors from rare but possible double execution, and + loudly warn the user whenever the problem is encountered (GH#63) + - Relax the 'self_result_object' argument check in the relationship + resolution codepath, restoring exotic uses of inflate_result + http://lists.scsys.co.uk/pipermail/dbix-class/2015-January/011876.html + - Fix updating multiple CLOB/BLOB columns on Oracle + - Fix exception on complex update/delete under a replicated setup + http://lists.scsys.co.uk/pipermail/dbix-class/2015-January/011903.html + - Fix uninitialized warnings on empty hashes passed to join/prefetch + https://github.com/vanstyn/RapidApp/commit/6f41f6e48 and + http://lists.scsys.co.uk/pipermail/dbix-class/2015-February/011921.html + - Fix hang in t/72pg.t when run against DBD::Pg 3.5.0. The ping() + implementation changes due to RT#100648 made an alarm() based + timeout lock-prone. + + * Misc + - Remove warning about potential side effects of RT#79576 (scheduled) + - Various doc improvements (GH#35, GH#62, GH#66, GH#70, GH#71, GH#72) + - Depend on newer Moo, to benefit from a safer runtime (RT#93004) + - Fix intermittent failures in the LeakTracer on 5.18+ + - Fix failures of t/54taint.t on Windows with spaces in the $^X + executable path (RT#101615) + +0.082810 2014-10-25 13:58 (UTC) + * Fixes + - Fix incorrect collapsing-parser source being generated in the + presence of unicode data among the collapse-points + - Fix endless loop on BareSourcelessResultClass->throw_exception(...) + + * Misc + - Depend on newer SQL::Abstract (fixing overly-aggressive parenthesis + opener: RT#99503) + - Depend on newer Moo, fixing some interoperability issues: + http://lists.scsys.co.uk/pipermail/dbix-class/2014-October/011787.html + +0.082801 2014-10-05 23:55 (UTC) + * Known Issues + - Passing large amounts of objects with stringification overload + directly to DBIx::Class may result in strange action at a distance + exceptions. More info (and a workaround description) can be found + under "Note" at https://metacpan.org/pod/SQL::Abstract#is_plain_value + - The relationship condition resolution fixes come with the side effect + of returning more complete data, tripping up *some* users of an + undocumented but widely used internal function. In particular + https://rt.cpan.org/Ticket/Display.html?id=91375#txn-1407239 + + * Notable Changes and Deprecations + - DBIC::FilterColumn now properly bypasses \'' and \[] literals, just + like the rest of DBIC + - DBIC::FilterColumn "from_storage" handler is now invoked on NULLs + returned from storage + - find() now throws an exception if some of the supplied values are + managed by DBIC::FilterColumn (RT#95054) + - Custom condition relationships are now invoked with a slightly + different signature (existing coderefs will continue to work) + - Add extra custom condition coderef attribute 'foreign_values' + to allow for proper reverse-relationship-like behavior + (i.e. $result->set_from_related($custom_rel, $foreign_result_object) + - When in a transaction, DBIC::Ordered now seamlesly handles result + objects that went out of sync with the storage (RT#96499) + - CDBICompat::columns() now supports adding columns through supplied + Class::DBI::Column instances (GH#52) + - Deprecate { col1 => col2 } expressions in manual {from} structures + (at some point of time manual {from} will be deprecated entirely) + + * Fixes + - Fix Resultset delete/update affecting *THE ENTIRE TABLE* in cases + of empty (due to conditions) resultsets with multi-column keys + - Fix on_connect_* not always firing in some cases - a race condition + existed between storage accessor setters and the determine_driver + routines, triggering a connection before the set-cycle is finished + - Fix collapse being ignored on single-origin selection (RT#95658) + - Fix incorrect behavior on custom result_class inflators altering + the amount of returned results + - Fix failure to detect stable order criteria when in iterator + mode of a has_many prefetch off a search_related chain + - Prevent erroneous database hit when accessing prefetched related + resultsets with no rows + - Proper exceptions on malformed relationship conditions (RT#92234) + - Fix incorrect handling of custom relationship conditions returning + SQLA literal expressions + - Fix long standing bug with populate() missing data from hashrefs with + different keysets: http://is.gd/2011_dbic_populate_gotcha (RT#92723) + - Fix multi-value literal populate not working with simplified bind + specifications + - Massively improve the implied resultset condition parsing - now all + applicable conditions within a resultset should be properly picked + up by create() and populate() + - Ensure definitive condition extractor handles bizarre corner cases + without bombing out (RT#93244) + - Fix set_column on non-native (+columns) selections (RT#86685) + - Fix set_inflated_column incorrectly handling \[] literals (GH#44) + - Ensure that setting a column to a literal invariably marks it dirty + - Fix copy() not working correctly with extra selections present + - Work around exception objects with broken string overloading in one + additional codepath (missed in 0.08260) + - Fix more inconsistencies of the quote_names attribute propagating + to SQL::Translator (partially RT#87731) + - Fix SQLT constraint naming when DBIC table names are fully qualified + (PR#48) + - Ensure ::Schema::Versioned connects only once by reusing the main + connection (GH#57) + - Fix inability to handle multiple consecutive transactions with + savepoints on DBD::SQLite < 1.39 + - Fix CDBICompat to match Class::DBI behavior handling non-result + blessed has_a (implicit deflate via stringification and inflate via + blind new) (GH#51) + + * Misc + - Ensure source metadata calls always take place on the result source + instance registered with the caller + - IFF DBIC_TRACE output defaults to STDERR we now silence the possible + wide-char warnings if the trace happens to contain unicode + +0.08270 2014-01-30 21:54 (PST) + * Fixes + - Fix 0.08260 regression in DBD::SQLite bound int handling. Inserted + data was not affected, but any function <=> integer comparison would + have failed (originally fixed way back in 0e773352) + - Fix failure to load DateTime formatter when connecting to Firebird + over ODBC + + * Misc + - All drivers based on ::Storage::DBI::Firebird::Common now return the + same sqlt_type value (affects ::DBI::Interbase, ::DBI::Firebird and + ::DBI::ODBC::Firebird) + +0.08260 2014-01-28 18:52 (UTC) + * New Features + - A new zero-to-DBIC style manual: DBIx::Class::Manual::QuickStart + + * Notable Changes and Deprecations + - Explicitly deprecate combination of distinct and selecting a + non-column via $rs->get_column() + + * Fixes + - More robust handling of circular relationship declarations by loading + foreign classes less frequently (should resolve issues like + http://lists.scsys.co.uk/pipermail/dbix-class/2013-June/011374.html) + Note that none of this is a manifestations of a DBIC bug, but rather + unexpected (but correct) behavior of load-order-dependent (hence + logically broken) Resultclass hierarchies. In order to deal with this + DBIC is scaling back a large number of sanity checks, which are to be + reintroduce pending a better framework for source registration + - Fix multiple edge cases of complex prefetch combining incorrectly + with correlated subquery selections + - Fix multiple edge cases stemming from interaction of a non-selecting + order_by specification and distinct and/or complex prefetch + - Fix unbound growth of a resultset during repeated execute/exhaust + cycles (GH#29) + - Work around (and be very vocal about the fact) when DBIC encounters + an exception object with broken string overloading + - Clarify ambiguous behavior of distinct when used with ResultSetColumn + i.e. $rs->search({}, { distinct => 1 })->get_column (...) + - Setting quote_names propagates to SQL::Translator when producing + SQLite DDL (it is one of the few producers *NOT* quoting by default) + - Fix incorrect binding of large integers on old versions of + DBD::SQLite (now DBIC simply always binds SQLite ints as BIGINT) + - Silence (harmless) warnings on recent DBD::SQLite versions, when + inserting/updating large integers on 32bit ivsize systems (RT#76395) + - Back out self-cleaning from DBIx::Class::Carp for the time being + (as a side effect fixes RT#86267) + - Fix incorrect internal use of implicit list context in copy() + - Fix 0.08250 regression in driver determination when DBI_DSN is used + - Tests no longer fail if $ENV{DBI_DSN} is set + - Throw clearer exception on ->new_related() with a non-existent + relationship + - Fix incorrect parethesis unroll with multicolumn in, (never noticed + before fixing false positive in SQLA::Test 1.77) + - Fix t/storage/replicated.t class loading problem + - Stop using the deprecated Class::MOP::load_class() + - Fix warning in t/54taint.t with explicitly unset PERL5LIB (RT#91972) + - Fix t/54taint.t failing under a local::lib with installed earlier + DBIC version (RT#92486) + + * Misc + - Massive incompatible change of ::BlockRunner internals (was never + documented as usable externally, this last set of changes settles + the design for proper documentation and opening up) + - Adjust exceptions in tests to accommodate changes in the upcoming + DBD::SQLite based on libsqlite 3.8.2 + - More robust lock file naming scheme - allow tests to work on exotic + MSWin32 filesystems (habitual offender being http://is.gd/iy5XVP) + - Better diagnostics when File::Spec->tmpdir gives us crap in testing + - Replace $row with $result in all docs to be consistent and to + clarify various return values + +0.08250 2013-04-29 22:00 (UTC) + * New Features / Changes + - Rewrite from scratch the result constructor codepath - many bugfixes + and performance improvements (the current codebase is now capable of + outperforming both DBIx::DataModel and Rose::DB::Object on some + workloads). Some notable benefits: + - Multiple has_many prefetch + - Partial prefetch - you now can select only columns you are + interested in, while preserving the collapse functionality + (collapse is now exposed as a first-class API attribute) + - Prefetch of resultsets with arbitrary order + (RT#54949, RT#74024, RT#74584) + - Prefetch no longer inserts right-side table order_by clauses + (massively helps the deficient MySQL optimizer) + - Prefetch with limit on right-side ordered resultsets now works + correctly (via aggregated grouping) + - No longer order the insides of a complex prefetch subquery, + unless required to satisfy a limit + - Stop erroneously considering order_by criteria from a join under + distinct => 1 (the distinct should apply to the main source only) + - Massively optimize codepath around ->cursor(), over 10x speedup + on some iterating workloads. + - Support standalone \[ $sql, $value ] in literal SQL with bind + specifications: \[ '? + ?', 42, 69 ] is now equivalent to + \[ '? + ?', [ {} => 42 ], [ {} => 69 ] ] + - Changing the result_class of a ResultSet in progress is now + explicitly forbidden. The behavior was undefined before, and + would result in wildly differing outcomes depending on $rs + attributes. + - Deprecate returning of prefetched 'filter' rels as part of + get_columns() and get_inflated_columns() data + - Invoking get_inflated_columns() no longer fires get_columns() but + instead retrieves data from individual non-inflatable columns via + get_column() + - Emit a warning on incorrect use of nullable columns within a + primary key + - Limited checks are performed on whether columns without declared + is_nullable => 1 metadata do in fact sometimes fetch NULLs from + the database (the check is currently very limited and is performed + only on resultset collapse when the alternative is rather worse) + + * Fixes + - Fix _dbi_attrs_for_bind() being called befor DBI has been loaded + (regression in 0.08210) + - Fix update/delete operations on resultsets *joining* the updated + table failing on MySQL. Resolves oversights in the fixes for + RT#81378 and RT#81897 + - Fix open cursors silently resetting when inherited across a fork + or a thread + - Properly support "MySQL-style" left-side group_by with prefetch + - Fix $grouped_rs->get_column($col)->func($func) producing incorrect + SQL (RT#81127) + - Stop Sybase ASE storage from generating invalid SQL in subselects + when a limit without offset is encountered + - Even more robust behavior of GenericSubQuery limit dialect + - Make sure deployment_statements() and cursor_class() are called on + a resolved storage subclass + + * Misc + - Fix tests failing due to unspecified resultset retrieval order + (test suite now will pass with newest SQLite libs) + +0.08210 2013-04-04 15:30 (UTC) * New Features / Changes - Officially deprecate the 'cols' and 'include_columns' resultset attributes - Remove ::Storage::DBI::sth() deprecated in 0.08191 * Fixes + - Work around a *critical* bug with potential for data loss in + DBD::SQLite - RT#79576 - Audit and correct potential bugs associated with braindead reuse of $1 on unsuccessful matches + - Fix incorrect warning/exception originator reported by carp*() and + throw_exception() 0.08209 2013-03-01 12:56 (UTC) * New Features / Changes