John Napiorkowski [Mon, 8 Jun 2009 21:32:00 +0000]
got the tests working again
John Napiorkowski [Mon, 8 Jun 2009 21:19:56 +0000]
added test for run againt storage method and share directory for stuff
John Napiorkowski [Mon, 8 Jun 2009 21:08:42 +0000]
more details in the AuthorCheck.pm to help newcomers understand why they are being threatened
Peter Rabbitson [Mon, 8 Jun 2009 09:36:56 +0000]
Simply order_by/_virtual_order_by handling
Peter Rabbitson [Sun, 7 Jun 2009 23:24:06 +0000]
Fix find_or_new/create to stop returning random rows when default value insert is requested
Peter Rabbitson [Sun, 7 Jun 2009 23:03:04 +0000]
Make podcoverage happy
Peter Rabbitson [Sun, 7 Jun 2009 22:59:21 +0000]
Add mysql empty insert SQL override
Make SQLAHacks parts loadable at runtime via ensure_class_loaded
Peter Rabbitson [Sun, 7 Jun 2009 21:36:43 +0000]
Make empty/default inserts use standard SQL
Peter Rabbitson [Sun, 7 Jun 2009 21:07:55 +0000]
Fix for mysql subquery problem
Peter Rabbitson [Sat, 6 Jun 2009 08:50:56 +0000]
Port another forgotten MC fix
Peter Rabbitson [Sat, 6 Jun 2009 08:47:47 +0000]
TODOified reentrancy counter - this shall be used to optimize MC some day
Peter Rabbitson [Sat, 6 Jun 2009 08:45:10 +0000]
rename MC test dir
Peter Rabbitson [Fri, 5 Jun 2009 17:25:56 +0000]
Factor out the grouped count subquery SELECTor so specific storage classes can override it
Trim attribute deletion lists - we work with _resolved_attrs - +X are already folded into X and the like
Peter Rabbitson [Fri, 5 Jun 2009 14:28:58 +0000]
This was dumb - of course I want a copy
Otherwise the next serach_related will work with a contaminated seen_joins
Peter Rabbitson [Thu, 4 Jun 2009 21:51:13 +0000]
It seems that this localisation can lead to problems
Unfortunately no test case as the trigger was buried deep within Reaction, I coulnd't figure out an isolation
Peter Rabbitson [Thu, 4 Jun 2009 20:54:28 +0000]
having without group_by is useless - thus do not supply a group_by on count, let things die
Peter Rabbitson [Thu, 4 Jun 2009 13:38:21 +0000]
Really delegate counting to the storage class - now we have either a grouped count or a regular one
Peter Rabbitson [Thu, 4 Jun 2009 09:52:18 +0000]
Delegate actual counting to the storage class
Peter Rabbitson [Thu, 4 Jun 2009 07:35:25 +0000]
Move relationship tests around
Peter Rabbitson [Wed, 3 Jun 2009 08:14:51 +0000]
Require a recent version of Date::Simple during CDBI tests
Peter Rabbitson [Mon, 1 Jun 2009 14:49:09 +0000]
Fix fallout from another botched merge (I suck, part 2)
Peter Rabbitson [Mon, 1 Jun 2009 14:41:49 +0000]
Merge 'order_by_refactor' into 'trunk'
Peter Rabbitson [Mon, 1 Jun 2009 10:24:41 +0000]
deploy-related pod fixes
Peter Rabbitson [Sun, 31 May 2009 14:07:18 +0000]
Last set of Makefile.PL optimizations
Peter Rabbitson [Sun, 31 May 2009 09:24:15 +0000]
There is a saner way to write out resources
Peter Rabbitson [Sun, 31 May 2009 07:42:35 +0000]
Throw away the makefile SQLite test - it served its purpose
Peter Rabbitson [Sun, 31 May 2009 07:37:37 +0000]
M::I 0.89 finally resolves all provlems with auto_install
Peter Rabbitson [Sat, 30 May 2009 17:46:09 +0000]
populate() fix and Changes
Peter Rabbitson [Sat, 30 May 2009 17:17:02 +0000]
Add changes
Peter Rabbitson [Sat, 30 May 2009 17:13:23 +0000]
Evil hack to make Carp::Clan work throughout SQLA as well
Peter Rabbitson [Sat, 30 May 2009 15:15:57 +0000]
Greatly simplify _order_by override to fallback on new SQLA
Peter Rabbitson [Sat, 30 May 2009 14:55:37 +0000]
Restructure bind tests
Peter Rabbitson [Sat, 30 May 2009 14:54:37 +0000]
TODOify some of the order with bind tests
Moritz Onken [Sat, 30 May 2009 09:06:54 +0000]
order_by tests
Peter Rabbitson [Sat, 30 May 2009 07:34:20 +0000]
Add better error reporting on bulk_insert (ash++)
Peter Rabbitson [Sat, 30 May 2009 05:40:24 +0000]
Patch + test for more informative exceptions on load_namespace a non-rs class
Peter Rabbitson [Thu, 28 May 2009 11:02:22 +0000]
Fix multiprefetch warning - we can now count properly
Peter Rabbitson [Thu, 28 May 2009 08:20:57 +0000]
Commit rather useless but already written mysql test extension
Peter Rabbitson [Tue, 26 May 2009 20:02:29 +0000]
Release 0.08103
Peter Rabbitson [Tue, 26 May 2009 19:28:49 +0000]
Attempt to reproduce reported mysql error (failed) - fixed another bug in ResultSetColumn along the way
Peter Rabbitson [Tue, 26 May 2009 18:49:49 +0000]
fix comments
Peter Rabbitson [Tue, 26 May 2009 15:20:57 +0000]
Minor fixes of the return value of rs->update/delete
Peter Rabbitson [Tue, 26 May 2009 14:36:55 +0000]
Fix test skip message
Peter Rabbitson [Tue, 26 May 2009 14:29:55 +0000]
Merge 'top_limit_altfix' into 'trunk'
Ash Berlin [Tue, 26 May 2009 14:17:44 +0000]
Fix 'timestamp with time zone' columns for IC::DT inflation
Peter Rabbitson [Tue, 26 May 2009 06:15:34 +0000]
TODOified test for RT#40701
Peter Rabbitson [Mon, 25 May 2009 17:42:02 +0000]
This method does not exist anymore
Tom Hukins [Mon, 25 May 2009 16:33:46 +0000]
Small documentation improvement: link to a documented method.
Peter Rabbitson [Mon, 25 May 2009 15:45:33 +0000]
Too much logic for no benefit - always populate _virtual_order_by
Peter Rabbitson [Mon, 25 May 2009 14:53:13 +0000]
Actually don't need this anymore
Peter Rabbitson [Mon, 25 May 2009 14:08:40 +0000]
Merge 'top_limit_tweaks' into 'top_limit_altfix'
Shoot another Top problem, move test from top_limit_tweaks branch and delete
Peter Rabbitson [Mon, 25 May 2009 09:24:32 +0000]
The Top limit emulation bundled with SQLA::Limit assumes that the limited resultset will be _always_ sorted. In order to fix this, I reimplemented _Top in SQLAHacks with a slight modification. Now the original order_by is passed to the outside of the nested select block, while order_up/down are calculated either based on the original order_by, or if one is not present an order by all PKs is attempted.
Since I do not have access to $rsrc in SQLA, I pass the list of PKs as an extra group_by hash entry. This appears to be rather safe, and besides we already pollute order_by with group_by and having (which seems to work rather well).
The only thing I am unsure about is the need for _gen_virtual_order(). Initially I was going to generate the pk list, only if we use the Top limit. Then it turned out there is no limit dialect before we connect, so I commented it out. Now all it does is check for a limit condition and returns the PK list. Is this necessary at all?
Peter Rabbitson [Mon, 25 May 2009 09:11:01 +0000]
Define how Top limit emulation should behave
Peter Rabbitson [Mon, 25 May 2009 07:47:01 +0000]
Forgotten podcoverage override
Peter Rabbitson [Mon, 25 May 2009 07:42:45 +0000]
Move the DB2 Limit syntax setting into the storage class
Peter Rabbitson [Mon, 25 May 2009 06:48:38 +0000]
Factor out the order_by sqlahacks resolver
Rafael Kitover [Sun, 24 May 2009 23:53:12 +0000]
added test to make sure only one connection to ODBC/MSSQL is made
Rafael Kitover [Sun, 24 May 2009 23:49:17 +0000]
fix double connect for ODBC/MSSQL
Peter Rabbitson [Sun, 24 May 2009 20:53:15 +0000]
Switch around inheritance of MSSQL drivers, remove some duplicate code
Peter Rabbitson [Sun, 24 May 2009 13:35:07 +0000]
Failing test about warnings triggered in SQLA::Limit when using a subquery
Peter Rabbitson [Sun, 24 May 2009 13:00:50 +0000]
eol adjustments
Peter Rabbitson [Sun, 24 May 2009 12:12:39 +0000]
Deprecate ::DBI::Sybase::MSSQL
Peter Rabbitson [Sun, 24 May 2009 09:43:32 +0000]
SUPER is so last century
Peter Rabbitson [Sun, 24 May 2009 09:42:01 +0000]
Make sure we do not clobber search attributes when using subqueries
Peter Rabbitson [Sun, 24 May 2009 08:47:03 +0000]
MSSQL through ODBC does not like unfinished statements - make sure we finish the scope identity retrieval
(This worked before because of the automatic retry on exception, essentially running any select after insert twice)
Peter Rabbitson [Sat, 23 May 2009 22:50:08 +0000]
Extend distinct deprecation tests and clarify warnings
Peter Rabbitson [Sat, 23 May 2009 22:47:40 +0000]
update changes
Peter Rabbitson [Sat, 23 May 2009 21:37:19 +0000]
Make sure MultiColumnIn quotes column names while munging literal sql
Tested by running t/resultset/update_delete against pg with quote char " and name sep .
Rafael Kitover [Sat, 23 May 2009 21:04:01 +0000]
minor clean up
Peter Rabbitson [Sat, 23 May 2009 20:59:17 +0000]
Extra test and count fixes for prefetch + distinct
Rafael Kitover [Sat, 23 May 2009 20:28:42 +0000]
revert last Sybase changes
Rafael Kitover [Sat, 23 May 2009 19:57:08 +0000]
hide internal Sybase classes from PAUSE
Rafael Kitover [Sat, 23 May 2009 19:46:11 +0000]
Sybase bindvar and IC::DT support
Peter Rabbitson [Sat, 23 May 2009 19:35:59 +0000]
Add storage component of multipk resultset update/delete for multicolumn IN capable rdbms
Switch Pg to that (tested via DBICTEST_DSN)
Peter Rabbitson [Sat, 23 May 2009 19:33:28 +0000]
Tests and test schema adjustments for resultset update/delete
Rafael Kitover [Sat, 23 May 2009 14:44:59 +0000]
Sybase autopk, and a test, no limit support yet
Peter Rabbitson [Sat, 23 May 2009 08:09:46 +0000]
Refactor rs_update_delete (too many methods for no reason)
Make mysql use _per_row_update_delete by default as this is the only thing it understands: http://dev.mysql.com/doc/refman/5.0/en/subquery-errors.html
Ash Berlin [Fri, 22 May 2009 16:54:00 +0000]
Make distinct calculate columns *after* prefetch has been resolved. Tests to come
Rafael Kitover [Fri, 22 May 2009 13:50:45 +0000]
::Replicated -- check for master rebless and reapply role if necessary
Peter Rabbitson [Fri, 22 May 2009 08:35:36 +0000]
duh
Peter Rabbitson [Fri, 22 May 2009 08:31:25 +0000]
Some attributes require a grouped count subquery, some - just a subquery. Differentiate properly
Peter Rabbitson [Fri, 22 May 2009 08:28:58 +0000]
Extend DB2 test
Rafael Kitover [Fri, 22 May 2009 06:09:05 +0000]
fix limit for DB2
Rafael Kitover [Fri, 22 May 2009 04:52:47 +0000]
comment addition
Rafael Kitover [Fri, 22 May 2009 02:54:12 +0000]
add some support for trying to determine storage driver without being connected
Peter Rabbitson [Thu, 21 May 2009 07:59:50 +0000]
Duh, no point of having ordered subqueries while counting, Debolaz++
Peter Rabbitson [Thu, 21 May 2009 07:45:52 +0000]
Clarify usage of _resolved_attrs by adding the explicit _resolved_attrs_copy
Clarify code in ResultSetColumn
Peter Rabbitson [Wed, 20 May 2009 21:47:05 +0000]
on_connect_do now accepts a single string like it does an arrayref (patch by prema)
Peter Rabbitson [Wed, 20 May 2009 21:00:13 +0000]
Switch as_query testing to direct specification in is_same_sql_bind
Peter Rabbitson [Wed, 20 May 2009 20:59:10 +0000]
POD fixes
Arthur Axel "fREW" Schmidt [Wed, 20 May 2009 16:03:55 +0000]
fix test for people with spaces in path
Justin Hunter [Wed, 20 May 2009 15:06:50 +0000]
fix for sql functions in group_by
Justin Hunter [Wed, 20 May 2009 15:04:17 +0000]
update to Changes
Peter Rabbitson [Wed, 20 May 2009 13:42:48 +0000]
More explicit attribute deletion/detection - a lof of this is probably redundant, as _resoved_attrs() should flatten them away (i.e. prefetch to collapse and distinct to group_by). Nevertheless keep them there for explcetness ( delete() is cheap )
Peter Rabbitson [Wed, 20 May 2009 13:32:27 +0000]
Make sure we always work with resolved rs attributes while counting (and test)
Peter Rabbitson [Wed, 20 May 2009 12:30:40 +0000]
Add explicit grouping for rs update/delete operations if the parameters warrant it (tests coming tonight)
Peter Rabbitson [Wed, 20 May 2009 10:51:05 +0000]
Bring back the old code from resolve_prefetch so ash's code can work
Add tests to make sure we throw it away in 09
Peter Rabbitson [Wed, 20 May 2009 10:44:54 +0000]
detect/purge having bind for compeleteness
Peter Rabbitson [Wed, 20 May 2009 07:03:08 +0000]
Make joined rs counts backwards compatible - we do not collapse a result exploded by a has_many join unless it is explicitly requested by distinct => 1, OR unless we have collapse set which implies prefetch
Rafael Kitover [Wed, 20 May 2009 03:58:36 +0000]
add test for distinct result of sql function