Release 1.82

[scpubgit/Q-Branch.git] / Changes

Revision history for SQL::Abstract

revision 1.82  2017-03-20
-------------------------
    - Add explicit dependency on Sub::Quote (GH#8)
    - Fix syntax errors in ORDER BY docs (GH#7)

revision 1.81_01  2017-02-28
----------------------------
    - Fix order clauses with bind parameters in ->where
    - Fix ->insert($table, \@values) with >26 values (RT#112684)
    - Teach ::Tree that ILIKE (PostgreSQL) and REGEXP (MySQL) are binary ops
    - Support for UPDATE ... RETURNING
    - Documentation improvements for ORDER BY

revision 1.81  2014-10-25
----------------------------
    - Fix overly-enthusiastic parenthesis unroller (RT#99503)

revision 1.80  2014-10-05
----------------------------
    - Fix erroneous behavior of is_literal_value($) wrt { -ident => ... }
    - Explicitly croak on top-level special ops (they didn't work anyway)

revision 1.79  2014-09-25
----------------------------
    - New exportable functions: is_literal_value($) and is_plain_value($)
    - New attribute 'escape_char' allowing for proper escape of quote_chars
      present in an identifier
    - Deprecate { "" => \... } constructs
    - Treat { -value => undef } as plain undef in all cases
    - Explicitly throw on { -ident => undef }

revision 1.78  2014-05-28
----------------------------
    - Fix parsing of binary ops to correctly take up only a single LHS
      element, instead of gobbling up the entire parse-to-date
    - Explicitly handle ROW_NUMBER() OVER as the snowflake-operator it is
    - Improve signatures/documentation of is_same_sql_bind / eq_sql_bind
    - Retire script/format-sql - the utility needs more work to be truly
      end-user convenient

revision 1.77  2014-01-17
----------------------------
    - Reintroduce { -not => undef } column operator (regression from 1.75)

revision 1.75  2013-12-27
----------------------------
    - *UPCOMING INCOMPATIBLE BUGFIX*: SQLA used to generate incorrect SQL
      on undef-containing lists fed to -in and -not_in. An exception will
      be raised for a while before properly fixing this, to avoid quiet
      but subtle changes to query results in production
    - Deprecate and warn when supplying an empty arrayref to like/not_like
      operators (likely to be removed before 2.0)

    - Warn when using an inequality operator with a multi-value array to
      arrive at what amounts to a 1=1 condition (no pre-2.0 plans to fix
      this behavior due to backwards comp concerns)
    - Fix false negative comparison of ORDER BY <function> ASC
    - More improvements of incorrect parsing (placeholder at end of list
      element)
    - Fix typos in POD and comments (RT#87776)
    - Augment -not_bool example with nesting (RT#89601)

revision 1.74  2013-06-04
----------------------------
    - Fix insufficient parenthesis unroll during operator comparison
    - 'ORDER BY foo' and 'ORDER BY foo ASC' are now considered equal
      by default (with a switch to reenable old behavior when necessary)
    - Change parser to not eagerly slurp RHS expressions it doesn't recognize

revision 1.73  2012-07-10
----------------------------
    - Fix parsing of ORDER BY foo + ?
    - Stop filling in placeholders in `format-sql` since it does not support
      passing values for them anyway
    - Fix parsing of NOT EXISTS
    - Fix over-eager parenthesis unrolling
    - Fix deep recursion warnings while parsing obnoxiously long sql statements
    - Fix incorrect comparison of malformed lists
    - Fix incorrect reporting of mismatch-members in SQLA::Test
    - Migrate the -ident operator from DBIC into SQLA
    - Migrate the -value operator from DBIC into SQLA

revision 1.72  2010-12-21
----------------------------
    - Extra checks of search arguments for possible SQL injection attacks
    - Remove excess parentheses in debug SQL
    - Fix parsing of foo.* in SQLA::Tree
    - Fix bindtype fail when using -between with arrayrefref literals
    - Add handling for NULL for -in
    - The -nest operator has entered semi-deprecated status and has been
      undocumented. Please do not use it in new code

revision 1.71  2010-11-09
----------------------------
    - Add EXECUTING for clarity of long running SQL
    - Add "squash_repeats" option to fix it such that repeated SQL gets ellided
      except for placeholders
    - Highlight transaction keywords
    - Highlight HAVING
    - Leave quotes from DBIC in bindargs
    - Add error checking on "profile" for SQLA::Tree
    - Hide bulk inserts from DBIx::Class
    - Fix missing doc (RT#62587)
    - Format functions in MySQL-friendly manner foo( ... ) vs foo ( ... )

revision 1.69  2010-10-22
----------------------------
    - Add quotes for populated placeholders and make the background
      magenta instead of cyan
    - Color and indent pagination keywords
    - Fix a silly bug which broke placeholder fill-in in DBIC
    - Installs format-sql to format SQL passed in over STDIN
    - Switch the tokenizer to precompiled regexes (massive speedup)
    - Rudimentary handling of quotes ( 'WHERE' vs WHERE )
    - Fix extended argument parsing by IN/BETWEEN
    - Add proper handling of lists (foo,bar,?)
    - Better handling of generic -function's during AST construction
    - Special handle IS NOT? NULL
    - Make sure unparse() does not destroy a passed in \@bindargs
    - Support ops with _'s in them (valid in Oracle)
    - Properly parse both types of default value inserts
    - Allow { -func => $val } as arguments to UPDATE

revision 1.68  2010-09-16
----------------------------
    - Document methods on Tree
    - Add affordances for color coding placeholders
    - Change ::Tree::whitespace to whitespace_keyword

revision 1.67_03  2010-09-11
----------------------------
    - Add docs for SQL::Abstract::Tree->new
    - correcty merge profile and parameters
    - added fill_in_placeholders option for excellent copy/pasta

revision 1.67_02  2010-09-08
----------------------------
    - rename DBIx::Class::Storage::PrettyPrinter to DBIx::Class::Storage::Debug::PrettyPrint
    - decreased a lot of indentation from ::Tree
    - cleaned up handling of newlines inside of parens

revision 1.67_01  2010-09-06
----------------------------
    - Add SQL::Abstract::Tree
    - Add unindexed DBIx::Class::Storage::PrettyPrinter
    - Better documentation of undef/NULL in where clause
    - Depend on bugfixed Module::Install (now again installs
      on old < 5.8.3 perls)

revision 1.67  2010-05-31 14:21 (UTC)
----------------------------
    - Fix SQL::Test failure when first chunk is an unrecognized
      literal
    - Generic -not operator tests
    - More columns-bindtype assertion checks

revision 1.66  2010-04-27 02:44 (UTC)
----------------------------
    - Optimized the quoting mechanism, winning nearly 10%
      speedup on repeatable sql generation

revision 1.65  2010-04-11 19:59 (UTC)
----------------------------
    - Rerelease last version to not include .svn files
      and grab MANIFEST.SKIP from DBIx::Class so it
      won't happen again

revision 1.64  2010-04-11 16:58 (UTC)
----------------------------
    - Fix multiple generic op handling regressions by
      reverting the auto-equality assumption (turned out
      to be a very very bad idea)

revision 1.63  2010-03-24 09:56 (UTC)
----------------------------
    - Add ILIKE to the core list of comparision ops

revision 1.62  2010-03-15 11:06 (UTC)
----------------------------
    - Fixed open outer parens for a multi-line literal
    - Allow recursively-nested column-functions in WHERE
    - Bumped minimum perl to 5.6.2 and changed tests to
      rely on core dependencies

revision 1.61  2010-02-05 16:28 (UTC)
----------------------------
    - Allow INSERT to take additional attributes
    - Support for INSERT ... RETURNING
    - Another iteration of SQL::Abstract::Test fixes and improvements

revision 1.60  2009-09-22 11:03 (UTC)
----------------------------
    - fix a well masked error in the sql-test tokenizer

revision 1.59  2009-09-22 08:39 (UTC)
----------------------------
    - fixed a couple of untrapped undefined warnings
    - allow -in/-between to accept literal sql in all logical
      variants - see POD for details
    - unroll multiple parenthesis around IN arguments to accomodate
      crappy databases

revision 1.58  2009-09-04 15:20 (UTC)
----------------------------
    - expanded the scope of -bool and -not_bool operators
    - added proper testing support

revision 1.57  2009-09-03 20:18 (UTC)
----------------------------
    - added -bool and -not_bool operators

revision 1.56  2009-05-30 16:31 (UTC)
----------------------------
    - support for \[$sql, @bind] in order_by clauses e.g.:
      { -desc => \['colA LIKE ?', 'somestring'] }

revision 1.55  2009-05-17 22:54 (UTC)
----------------------------
    - make sure that sql generation does not mutate the supplied
      where condition structure

revision 1.54  2009-05-07 17:23 (UTC)
----------------------------
    - allow special_operators to take both code refs and method names
      (makes it possible to properly subclass the builtin ones)

revision 1.53  2009-04-30 14:58 (UTC)
----------------------------
    - make sure hash keys are sorted in all search sub-conditions
    - switch installer from EU::MM to M::I

revision 1.52  2009-04-28 23:14 (UTC)
----------------------------
    - allow -between to handle [\"", \""] and \["", @bind]
    - allow order_by to handle -asc|desc => [qw/colA colB/] (artifact from DBIx::Class)
    - more tests and clearing up of some corner cases
    - t/10test.t does not run by default (developer only, too cpu intensive)

----------------------------
revision 1.51  2009-03-28 10:00 (UTC)
    - fixed behavior of [-and => ... ] depending on the current
      condition scope. This introduces backwards comp with 1.24

----------------------------
revision 1.50  2009-03-10 12:30 (UTC)
    - fixed the problem with values() not behaving the same as the rest of the code (RT#43483)
    - fixed interjecting arrayrefref into a where clause
    - added value-only insert test with a literal SQL snippet
    - cleanup and enhancement of t/03values.t
    - better handling of borked SQL in tests
    - deal properly with parentheses in is_same_sql_bind()
    - fixed test subs (is_same_*) in SQL::Abstract::Test to return the correct test value
    - do not version MANIFEST

    Version 1.50 was a major internal refactoring of SQL::Abstract.
    Great care has been taken to preserve the published behavior
    documented in previous versions in the 1.* family; however,
    some features that were previously undocumented, or behaved.
    differently from the documentation, had to be changed in order
    to clarify the semantics. Hence, client code that was relying
    on some dark areas of SQL::Abstract v1.* might behave differently
    in v1.50.

----------------------------
revision 1.49_04  2009-03-03
    - add support for a [\%column_meta => value] bind value format

----------------------------
revision 1.49_03  2009-02-17
    - clarify syntax of \['...', @bind] when used with a bindtype
      of 'columns'

----------------------------
revision 1.49_02  2009-02-16
    - added an AST-aware SQL::Abstract::Test library for sql syntax tests
    - vastly expanded test coverage
    - support for the { operator => \'...'|\['...', @bind] } syntax
      allowing to embed arbitrary operators on the LHS
    - fixed multiple regressions wrt DBIx::Class

----------------------------
revision 1.49_01  2009-02-11
    - support for literal SQL through the [$sql, bind] syntax.
    - added -nest1, -nest2 or -nest_1, -nest_2, ...
    - optional support for array datatypes
    - defensive programming : check arguments to functions/methods
    - fixed bug with global logic of -and/-or (no side-effects any more)
    - changed logic for distributing an op over arrayrefs
    - fixed semantics of  _bindtype on array args
    - dropped the _anoncopy of the %where tree. No longer necessary.
    - dropped the _modlogic function
    - Make col => [] and col => {$op => [] } DTRT or die instead of generating
      broken SQL. Added tests for this.
    - Added { -desc => 'column' } order by support
    - Tiny "$_"-related fix for { -desc => 'columns'} order by support
      tests + docs

----------------------------
revision 1.20
date: 2005/08/18 18:41:58;  author: nwiger;  state: Exp;  lines: +104 -50
- added patch from Dan Kubb enabling quote_char and name_sep options
- added patch from Andy Grundman to enhance _anoncopy for deep refs
----------------------------
revision 1.19
date: 2005/04/29 18:20:30;  author: nwiger;  state: Exp;  lines: +34 -20
added _anoncopy to prevent destroying original; updated docs
----------------------------
revision 1.18
date: 2005/03/07 20:14:12;  author: nwiger;  state: Exp;  lines: +201 -65
added support for -and, -or, and -nest; see docs for details
----------------------------
revision 1.17
date: 2004/08/25 20:11:27;  author: nwiger;  state: Exp;  lines: +58 -46
added patch from Eric Kolve to iterate over all hashref elements
----------------------------
revision 1.16
date: 2004/06/10 17:20:01;  author: nwiger;  state: Exp;  lines: +178 -12
added bindtype param to allow this to work with Orasuck 9+
----------------------------
revision 1.15
date: 2003/11/05 23:40:40;  author: nwiger;  state: Exp;  lines: +18 -6
several bugfixes, including _convert being applied wrong and
the edge case field => { '!=', [qw/this that/] } not working
----------------------------
revision 1.14
date: 2003/11/04 21:20:33;  author: nwiger;  state: Exp;  lines: +115 -34
added patch from Philip Collins, and also added 'convert' option
----------------------------
revision 1.13
date: 2003/05/21 17:22:29;  author: nwiger;  state: Exp;  lines: +230 -74
added "IN" and "BETWEEN" operator support, as well as "NOT"
modified where() to support ORDER BY, and fixed some bugs too
added PERFORMANCE and FORMBUILDER doc sections
fixed several bugs in _recurse_where(), it now works as expected
added test suite, many thanks to Chas Owens
modified all hash access to return keys sorted, to allow cached queries
----------------------------
revision 1.12
date: 2003/05/08 20:10:56;  author: nwiger;  state: Exp;  lines: +181 -96
1.11 interim checking; major bugfixes and order_by, 1.12 will go to CPAN
----------------------------
revision 1.11
date: 2003/05/02 00:07:30;  author: nwiger;  state: Exp;  lines: +52 -12
many minor enhancements to add querying flexibility
----------------------------
revision 1.10
date: 2002/09/27 18:06:25;  author: nwiger;  state: Exp;  lines: +6 -2
added precatch for messed up where string
----------------------------
revision 1.9
date: 2002/08/29 18:04:35;  author: nwiger;  state: Exp;  lines: +4 -3
CPAN
----------------------------
revision 1.8
date: 2001/11/07 22:18:12;  author: nwiger;  state: Exp;  lines: +31 -14
added embedded SCALAR ref capability to insert() and update()
----------------------------
revision 1.7
date: 2001/11/07 01:23:28;  author: nwiger;  state: Exp;  lines: +3 -3
damn uninit warning
----------------------------
revision 1.6
date: 2001/11/06 21:09:44;  author: nwiger;  state: Exp;  lines: +14 -6
oops, had to actually *implement* the order by for select()!
----------------------------
revision 1.5
date: 2001/11/06 03:13:16;  author: nwiger;  state: Exp;  lines: +43 -4
lots of docs
----------------------------
revision 1.4
date: 2001/11/06 03:07:42;  author: nwiger;  state: Exp;  lines: +16 -7
added extra layer of ()'s to ensure correct semantics on AND
----------------------------
revision 1.3
date: 2001/11/06 01:16:31;  author: nwiger;  state: Exp;  lines: +11 -10
updated all statements so that they use wantarray to just return SQL if asked
----------------------------
revision 1.2
date: 2001/10/26 22:23:46;  author: nwiger;  state: Exp;  lines: +112 -15
added scalar ref for SQL verbatim in where, fixed bugs, array ref, docs
----------------------------
revision 1.1
date: 2001/10/24 00:26:43;  author: nwiger;  state: Exp;
Initial revision