sqlobject-cvs — CVS notification

[SQL-CVS] r4584 - in SQLObject/trunk/sqlobject: mssql sybase

[<sub...@co...>]

Author: phd
Date: Sat Apr 20 04:24:27 2013
New Revision: 4584

Log:
print => printDebug to avoid problems with unicode

Modified:
   SQLObject/trunk/sqlobject/mssql/mssqlconnection.py
   SQLObject/trunk/sqlobject/sybase/sybaseconnection.py

Modified: SQLObject/trunk/sqlobject/mssql/mssqlconnection.py
==============================================================================
--- SQLObject/trunk/sqlobject/mssql/mssqlconnection.py	Wed Apr 10 14:25:03 2013	(r4583)
+++ SQLObject/trunk/sqlobject/mssql/mssqlconnection.py	Sat Apr 20 04:24:27 2013	(r4584)
@@ -140,7 +140,7 @@
 
         q = self._insertSQL(table, names, values)
         if self.debug:
-            print 'QueryIns: %s' % q
+            self.printDebug(conn, q, 'QueryIns')
         c.execute(q)
         if has_identity:
             c.execute('SET IDENTITY_INSERT %s OFF' % table)

Modified: SQLObject/trunk/sqlobject/sybase/sybaseconnection.py
==============================================================================
--- SQLObject/trunk/sqlobject/sybase/sybaseconnection.py	Wed Apr 10 14:25:03 2013	(r4583)
+++ SQLObject/trunk/sqlobject/sybase/sybaseconnection.py	Sat Apr 20 04:24:27 2013	(r4584)
@@ -83,7 +83,7 @@
 
         q = self._insertSQL(table, names, values)
         if self.debug:
-            print 'QueryIns: %s' % q
+            self.printDebug(conn, q, 'QueryIns')
         c.execute(q)
         if has_identity and identity_insert_on:
             c.execute('SET IDENTITY_INSERT %s OFF' % table)

[SQL-CVS] r4582 - SQLObject/trunk/sqlobject/tests

[<sub...@co...>]

Author: phd
Date: Sun Feb 10 11:50:54 2013
New Revision: 4582

Log:
Made the test works in all databases

Modified:
   SQLObject/trunk/sqlobject/tests/test_identity.py

Modified: SQLObject/trunk/sqlobject/tests/test_identity.py
==============================================================================
--- SQLObject/trunk/sqlobject/tests/test_identity.py	Tue Feb  5 13:23:32 2013	(r4581)
+++ SQLObject/trunk/sqlobject/tests/test_identity.py	Sun Feb 10 11:50:54 2013	(r4582)
@@ -9,10 +9,8 @@
     n = IntCol()
 
 def test_identity():
-    if getConnection().dbName != "mssql":
-        return
-
-    # create table
+    # (re)create table
+    TestIdentity.dropTable(connection=getConnection(), ifExists=True)
     setupClass(TestIdentity)
 
     # insert without giving identity
@@ -22,7 +20,7 @@
     assert(i1get.n == 100)
 
     # insert while giving identity
-    i1 = TestIdentity(id=2, n=200)
+    i2 = TestIdentity(id=2, n=200)
     # verify result
     i2get = TestIdentity.get(2)
     assert(i2get.n == 200)

[SQL-CVS] r4581 - in SQLObject/trunk: docs sqlobject/mssql sqlobject/tests

[<sub...@co...>]

Author: phd
Date: Tue Feb  5 13:23:32 2013
New Revision: 4581

Log:
Changed the way to get if the table has identity in MS SQL

Added:
   SQLObject/trunk/sqlobject/tests/test_identity.py
Modified:
   SQLObject/trunk/docs/Authors.txt
   SQLObject/trunk/docs/News.txt
   SQLObject/trunk/sqlobject/mssql/mssqlconnection.py

Modified: SQLObject/trunk/docs/Authors.txt
==============================================================================
--- SQLObject/trunk/docs/Authors.txt	Sun Feb  3 04:23:15 2013	(r4580)
+++ SQLObject/trunk/docs/Authors.txt	Tue Feb  5 13:23:32 2013	(r4581)
@@ -26,6 +26,7 @@
 * Daniel Fetchinson <fetchinson at googlemail.com>
 * Neil Muller <drnlmuller+sqlobject at gmail.com>
 * Petr Jakes <petr.jakes at tpc.cz>
+* Andrew Ziem <ahz001 at gmail.com>
 * Oleg Broytman <ph...@ph...>
 
 .. image:: http://sflogo.sourceforge.net/sflogo.php?group_id=74338&type=10

Modified: SQLObject/trunk/docs/News.txt
==============================================================================
--- SQLObject/trunk/docs/News.txt	Sun Feb  3 04:23:15 2013	(r4580)
+++ SQLObject/trunk/docs/News.txt	Tue Feb  5 13:23:32 2013	(r4581)
@@ -18,7 +18,9 @@
 
 * Optimization in PostgresConnection: use INSERT...RETURNING id
   to get the autoincremented id in one query instead of two
-  (SELECT id + INSERT).
+  (INSERT + SELECT id).
+
+* Changed the way to get if the table has identity in MS SQL.
 
 SQLObject 1.3.2
 ===============

Modified: SQLObject/trunk/sqlobject/mssql/mssqlconnection.py
==============================================================================
--- SQLObject/trunk/sqlobject/mssql/mssqlconnection.py	Sun Feb  3 04:23:15 2013	(r4580)
+++ SQLObject/trunk/sqlobject/mssql/mssqlconnection.py	Tue Feb  5 13:23:32 2013	(r4581)
@@ -100,13 +100,10 @@
         return con
 
     HAS_IDENTITY = """
-       SELECT col.name, col.status, obj.name
-       FROM syscolumns col
-       JOIN sysobjects obj
-       ON obj.id = col.id
-       WHERE obj.name = '%s'
-       and col.autoval is not null
-
+       select 1
+       from INFORMATION_SCHEMA.COLUMNS
+       where TABLE_NAME = '%s'
+       and COLUMNPROPERTY(object_id(TABLE_NAME), COLUMN_NAME, 'IsIdentity') = 1
     """
     def _hasIdentity(self, conn, table):
         query = self.HAS_IDENTITY % table

Added: SQLObject/trunk/sqlobject/tests/test_identity.py
==============================================================================
--- /dev/null	00:00:00 1970	(empty, because file is newly added)
+++ SQLObject/trunk/sqlobject/tests/test_identity.py	Tue Feb  5 13:23:32 2013	(r4581)
@@ -0,0 +1,28 @@
+from sqlobject import *
+from sqlobject.tests.dbtest import *
+
+########################################
+## Identity (MS SQL)
+########################################
+
+class TestIdentity(SQLObject):
+    n = IntCol()
+
+def test_identity():
+    if getConnection().dbName != "mssql":
+        return
+
+    # create table
+    setupClass(TestIdentity)
+
+    # insert without giving identity
+    i1 = TestIdentity(n=100)
+    # verify result
+    i1get = TestIdentity.get(1)
+    assert(i1get.n == 100)
+
+    # insert while giving identity
+    i1 = TestIdentity(id=2, n=200)
+    # verify result
+    i2get = TestIdentity.get(2)
+    assert(i2get.n == 200)

[SQL-CVS] r4580 - SQLObject/trunk/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:23:15 2013
New Revision: 4580

Log:
Merged revision 4578 from branch 1.2: corrected grammar

Modified:
   SQLObject/trunk/docs/SQLObject.txt

Modified: SQLObject/trunk/docs/SQLObject.txt
==============================================================================
--- SQLObject/trunk/docs/SQLObject.txt	Sun Feb  3 04:22:51 2013	(r4579)
+++ SQLObject/trunk/docs/SQLObject.txt	Sun Feb  3 04:23:15 2013	(r4580)
@@ -772,8 +772,9 @@
 
 .. _UnicodeCol: `Column Types`_
 
-The following attributes provide introspection but should not be set or
-directly - see `Runtime Column and Join Changes`_ for dynamically modifying these class elements.
+The following attributes provide introspection but should not be set directly -
+see `Runtime Column and Join Changes`_ for dynamically modifying these class
+elements.
 
 `columns`:
    A dictionary of ``{columnName: anSOColInstance}``.  You can get

[SQL-CVS] r4579 - SQLObject/branches/1.3/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:22:51 2013
New Revision: 4579

Log:
Merged revision 4578 from branch 1.2: corrected grammar

Modified:
   SQLObject/branches/1.3/docs/SQLObject.txt

Modified: SQLObject/branches/1.3/docs/SQLObject.txt
==============================================================================
--- SQLObject/branches/1.3/docs/SQLObject.txt	Sun Feb  3 04:21:36 2013	(r4578)
+++ SQLObject/branches/1.3/docs/SQLObject.txt	Sun Feb  3 04:22:51 2013	(r4579)
@@ -772,8 +772,9 @@
 
 .. _UnicodeCol: `Column Types`_
 
-The following attributes provide introspection but should not be set or
-directly - see `Runtime Column and Join Changes`_ for dynamically modifying these class elements.
+The following attributes provide introspection but should not be set directly -
+see `Runtime Column and Join Changes`_ for dynamically modifying these class
+elements.
 
 `columns`:
    A dictionary of ``{columnName: anSOColInstance}``.  You can get

[SQL-CVS] r4578 - SQLObject/branches/1.2/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:21:36 2013
New Revision: 4578

Log:
Corrected grammar

Modified:
   SQLObject/branches/1.2/docs/SQLObject.txt

Modified: SQLObject/branches/1.2/docs/SQLObject.txt
==============================================================================
--- SQLObject/branches/1.2/docs/SQLObject.txt	Sun Feb  3 04:17:07 2013	(r4577)
+++ SQLObject/branches/1.2/docs/SQLObject.txt	Sun Feb  3 04:21:36 2013	(r4578)
@@ -772,8 +772,9 @@
 
 .. _UnicodeCol: `Column Types`_
 
-The following attributes provide introspection but should not be set or
-directly - see `Runtime Column and Join Changes`_ for dynamically modifying these class elements.
+The following attributes provide introspection but should not be set directly -
+see `Runtime Column and Join Changes`_ for dynamically modifying these class
+elements.
 
 `columns`:
    A dictionary of ``{columnName: anSOColInstance}``.  You can get

[SQL-CVS] r4577 - SQLObject/trunk/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:17:07 2013
New Revision: 4577

Log:
Merged revision 4575 from branch 1.2: CodeSpeak.net has stopped hosting code

Modified:
   SQLObject/trunk/docs/DeveloperGuide.txt
   SQLObject/trunk/docs/links.txt

Modified: SQLObject/trunk/docs/DeveloperGuide.txt
==============================================================================
--- SQLObject/trunk/docs/DeveloperGuide.txt	Sun Feb  3 04:16:36 2013	(r4576)
+++ SQLObject/trunk/docs/DeveloperGuide.txt	Sun Feb  3 04:17:07 2013	(r4577)
@@ -144,12 +144,11 @@
 Tests are important.  Tests keep everything from falling apart.  All
 new additions should have tests.
 
-Testing uses `py.test`__, an alternative to ``unittest``.  It is
-available via subversion at http://codespeak.net/svn/py/dist.  Read
-its `getting started`_ document for more.
+Testing uses py.test, an alternative to ``unittest``.  It is available at
+http://pytest.org/ and http://pypi.python.org/pypi/pytest.  Read its `getting
+started`_ document for more.
 
-.. __: http://codespeak.net/py/current/doc/test.html
-.. _getting started: http://codespeak.net/py/dist/test.html#starting-point-py-test-command-line-tool
+.. _getting started: http://pytest.org/latest/getting-started.html
 
 To actually run the test, you have to give it a database to connect to.
 You do so with the option ``-D``. You can either give a complete URI or
@@ -175,7 +174,7 @@
 ``inserts`` function that can create instances for you, and a couple
 miscellaneous functions.
 
-.. _raises: http://codespeak.net/py/current/doc/test.html#id4
+.. _raises: http://pytest.org/latest/assert.html#assertions-about-expected-exceptions
 
 
 If you submit a patch or implement a feature without a test, I'll be

Modified: SQLObject/trunk/docs/links.txt
==============================================================================
--- SQLObject/trunk/docs/links.txt	Sun Feb  3 04:16:36 2013	(r4576)
+++ SQLObject/trunk/docs/links.txt	Sun Feb  3 04:17:07 2013	(r4577)
@@ -29,9 +29,6 @@
 Open Source Projects
 --------------------
 
-* `sqlos Zope 3 SQLObject integration
-  <http://codespeak.net/z3/sqlos/>`_.
-
 * `TurboGears <http://turbogears.org/>`_ Web Framework: also integrates
   with SQLObject.
 

[SQL-CVS] r4576 - SQLObject/branches/1.3/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:16:36 2013
New Revision: 4576

Log:
Merged revision 4575 from branch 1.2: CodeSpeak.net has stopped hosting code

Modified:
   SQLObject/branches/1.3/docs/DeveloperGuide.txt
   SQLObject/branches/1.3/docs/links.txt

Modified: SQLObject/branches/1.3/docs/DeveloperGuide.txt
==============================================================================
--- SQLObject/branches/1.3/docs/DeveloperGuide.txt	Sun Feb  3 04:12:43 2013	(r4575)
+++ SQLObject/branches/1.3/docs/DeveloperGuide.txt	Sun Feb  3 04:16:36 2013	(r4576)
@@ -144,12 +144,11 @@
 Tests are important.  Tests keep everything from falling apart.  All
 new additions should have tests.
 
-Testing uses `py.test`__, an alternative to ``unittest``.  It is
-available via subversion at http://codespeak.net/svn/py/dist.  Read
-its `getting started`_ document for more.
+Testing uses py.test, an alternative to ``unittest``.  It is available at
+http://pytest.org/ and http://pypi.python.org/pypi/pytest.  Read its `getting
+started`_ document for more.
 
-.. __: http://codespeak.net/py/current/doc/test.html
-.. _getting started: http://codespeak.net/py/dist/test.html#starting-point-py-test-command-line-tool
+.. _getting started: http://pytest.org/latest/getting-started.html
 
 To actually run the test, you have to give it a database to connect to.
 You do so with the option ``-D``. You can either give a complete URI or
@@ -175,7 +174,7 @@
 ``inserts`` function that can create instances for you, and a couple
 miscellaneous functions.
 
-.. _raises: http://codespeak.net/py/current/doc/test.html#id4
+.. _raises: http://pytest.org/latest/assert.html#assertions-about-expected-exceptions
 
 
 If you submit a patch or implement a feature without a test, I'll be

Modified: SQLObject/branches/1.3/docs/links.txt
==============================================================================
--- SQLObject/branches/1.3/docs/links.txt	Sun Feb  3 04:12:43 2013	(r4575)
+++ SQLObject/branches/1.3/docs/links.txt	Sun Feb  3 04:16:36 2013	(r4576)
@@ -29,9 +29,6 @@
 Open Source Projects
 --------------------
 
-* `sqlos Zope 3 SQLObject integration
-  <http://codespeak.net/z3/sqlos/>`_.
-
 * `TurboGears <http://turbogears.org/>`_ Web Framework: also integrates
   with SQLObject.
 

[SQL-CVS] r4575 - SQLObject/branches/1.2/docs

[<sub...@co...>]

Author: phd
Date: Sun Feb  3 04:12:43 2013
New Revision: 4575

Log:
CodeSpeak.net has stopped hosting code projects

Py.test got its own domain pytest.org. SQLOS died and disappeared from the net.

Modified:
   SQLObject/branches/1.2/docs/DeveloperGuide.txt
   SQLObject/branches/1.2/docs/links.txt

Modified: SQLObject/branches/1.2/docs/DeveloperGuide.txt
==============================================================================
--- SQLObject/branches/1.2/docs/DeveloperGuide.txt	Sat Jan 19 11:19:02 2013	(r4574)
+++ SQLObject/branches/1.2/docs/DeveloperGuide.txt	Sun Feb  3 04:12:43 2013	(r4575)
@@ -144,12 +144,11 @@
 Tests are important.  Tests keep everything from falling apart.  All
 new additions should have tests.
 
-Testing uses `py.test`__, an alternative to ``unittest``.  It is
-available via subversion at http://codespeak.net/svn/py/dist.  Read
-its `getting started`_ document for more.
+Testing uses py.test, an alternative to ``unittest``.  It is available at
+http://pytest.org/ and http://pypi.python.org/pypi/pytest.  Read its `getting
+started`_ document for more.
 
-.. __: http://codespeak.net/py/current/doc/test.html
-.. _getting started: http://codespeak.net/py/dist/test.html#starting-point-py-test-command-line-tool
+.. _getting started: http://pytest.org/latest/getting-started.html
 
 To actually run the test, you have to give it a database to connect to.
 You do so with the option ``-D``. You can either give a complete URI or
@@ -175,7 +174,7 @@
 ``inserts`` function that can create instances for you, and a couple
 miscellaneous functions.
 
-.. _raises: http://codespeak.net/py/current/doc/test.html#id4
+.. _raises: http://pytest.org/latest/assert.html#assertions-about-expected-exceptions
 
 
 If you submit a patch or implement a feature without a test, I'll be

Modified: SQLObject/branches/1.2/docs/links.txt
==============================================================================
--- SQLObject/branches/1.2/docs/links.txt	Sat Jan 19 11:19:02 2013	(r4574)
+++ SQLObject/branches/1.2/docs/links.txt	Sun Feb  3 04:12:43 2013	(r4575)
@@ -29,9 +29,6 @@
 Open Source Projects
 --------------------
 
-* `sqlos Zope 3 SQLObject integration
-  <http://codespeak.net/z3/sqlos/>`_.
-
 * `TurboGears <http://turbogears.org/>`_ Web Framework: also integrates
   with SQLObject.
 

[SQL-CVS] r4571 - SQLObject/branches/1.2/sqlobject

[<sub...@co...>]

Author: phd
Date: Sat Jan 19 11:06:10 2013
New Revision: 4571

Log:
Fixed a problem in test_paste - connection is None

Modified:
   SQLObject/branches/1.2/sqlobject/col.py

Modified: SQLObject/branches/1.2/sqlobject/col.py
==============================================================================
--- SQLObject/branches/1.2/sqlobject/col.py	Sat Dec 29 13:10:52 2012	(r4570)
+++ SQLObject/branches/1.2/sqlobject/col.py	Sat Jan 19 11:06:10 2013	(r4571)
@@ -505,12 +505,12 @@
             return None
         try:
             connection = state.connection or state.soObject._connection
+            binaryType = connection._binaryType
         except AttributeError:
             dbEncoding = "ascii"
             binaryType = type(None) # Just a simple workaround
         else:
             dbEncoding = getattr(connection, "dbEncoding", None) or "ascii"
-            binaryType = connection._binaryType
         if isinstance(value, unicode):
             return value.encode(dbEncoding)
         if self.dataType and isinstance(value, self.dataType):

[SQL-CVS] r4573 - SQLObject/trunk/sqlobject

[<sub...@co...>]

Author: phd
Date: Sat Jan 19 11:10:09 2013
New Revision: 4573

Log:
Merged revision 4571 from branch 1.2:
fixed a problem in test_paste - connection is None

Modified:
   SQLObject/trunk/sqlobject/col.py

Modified: SQLObject/trunk/sqlobject/col.py
==============================================================================
--- SQLObject/trunk/sqlobject/col.py	Sat Jan 19 11:09:27 2013	(r4572)
+++ SQLObject/trunk/sqlobject/col.py	Sat Jan 19 11:10:09 2013	(r4573)
@@ -505,12 +505,12 @@
             return None
         try:
             connection = state.connection or state.soObject._connection
+            binaryType = connection._binaryType
         except AttributeError:
             dbEncoding = "ascii"
             binaryType = type(None) # Just a simple workaround
         else:
             dbEncoding = getattr(connection, "dbEncoding", None) or "ascii"
-            binaryType = connection._binaryType
         if isinstance(value, unicode):
             return value.encode(dbEncoding)
         if self.dataType and isinstance(value, self.dataType):

[SQL-CVS] r4572 - SQLObject/branches/1.3/sqlobject

[<sub...@co...>]

Author: phd
Date: Sat Jan 19 11:09:27 2013
New Revision: 4572

Log:
Merged revision 4571 from branch 1.2:
fixed a problem in test_paste - connection is None

Modified:
   SQLObject/branches/1.3/sqlobject/col.py

Modified: SQLObject/branches/1.3/sqlobject/col.py
==============================================================================
--- SQLObject/branches/1.3/sqlobject/col.py	Sat Jan 19 11:06:10 2013	(r4571)
+++ SQLObject/branches/1.3/sqlobject/col.py	Sat Jan 19 11:09:27 2013	(r4572)
@@ -505,12 +505,12 @@
             return None
         try:
             connection = state.connection or state.soObject._connection
+            binaryType = connection._binaryType
         except AttributeError:
             dbEncoding = "ascii"
             binaryType = type(None) # Just a simple workaround
         else:
             dbEncoding = getattr(connection, "dbEncoding", None) or "ascii"
-            binaryType = connection._binaryType
         if isinstance(value, unicode):
             return value.encode(dbEncoding)
         if self.dataType and isinstance(value, self.dataType):

[SQL-CVS] r4574 - in SQLObject/trunk: docs sqlobject/postgres

[<sub...@co...>]

Author: phd
Date: Sat Jan 19 11:19:02 2013
New Revision: 4574

Log:
Use INSERT...RETURNING id

Optimization in PostgresConnection: use INSERT...RETURNING id
to get the autoincremented id in one query instead of two
(SELECT id + INSERT).
The feature requires PostgreSQL 8.2+.

Modified:
   SQLObject/trunk/docs/News.txt
   SQLObject/trunk/sqlobject/postgres/pgconnection.py

Modified: SQLObject/trunk/docs/News.txt
==============================================================================
--- SQLObject/trunk/docs/News.txt	Sat Jan 19 11:10:09 2013	(r4573)
+++ SQLObject/trunk/docs/News.txt	Sat Jan 19 11:19:02 2013	(r4574)
@@ -10,6 +10,16 @@
 SQLObject (trunk)
 =================
 
+Features & Interface
+--------------------
+
+* Support for PostgreSQL 8.1 is dropped. The minimal supported version of
+  PostgreSQL is 8.2 now.
+
+* Optimization in PostgresConnection: use INSERT...RETURNING id
+  to get the autoincremented id in one query instead of two
+  (SELECT id + INSERT).
+
 SQLObject 1.3.2
 ===============
 

Modified: SQLObject/trunk/sqlobject/postgres/pgconnection.py
==============================================================================
--- SQLObject/trunk/sqlobject/postgres/pgconnection.py	Sat Jan 19 11:10:09 2013	(r4573)
+++ SQLObject/trunk/sqlobject/postgres/pgconnection.py	Sat Jan 19 11:19:02 2013	(r4574)
@@ -189,15 +189,20 @@
         sequenceName = soInstance.sqlmeta.idSequence or \
                                '%s_%s_seq' % (table, idName)
         c = conn.cursor()
+        if id is not None:
+            names = [idName] + names
+            values = [id] + values
+        if names and values:
+            q = self._insertSQL(table, names, values)
+        else:
+            q = "INSERT INTO %s DEFAULT VALUES" % table
         if id is None:
-            self._executeRetry(conn, c, "SELECT NEXTVAL('%s')" % sequenceName)
-            id = c.fetchone()[0]
-        names = [idName] + names
-        values = [id] + values
-        q = self._insertSQL(table, names, values)
+            q += " RETURNING " + idName
         if self.debug:
             self.printDebug(conn, q, 'QueryIns')
         self._executeRetry(conn, c, q)
+        if id is None:
+            id = c.fetchone()[0]
         if self.debugOutput:
             self.printDebug(conn, id, 'QueryIns', 'result')
         return id

[SQL-CVS] [ sqlobject-Patches-3601405 ] Use "insert ... returning id" with postgres

[SourceForge.net <no...@so...>]

Patches item #3601405, was opened at 2013-01-18 10:39
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3601405&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Ken Lalonde (kenlalonde)
>Assigned to: Oleg Broytman (phd)
Summary: Use "insert ... returning id" with postgres

Initial Comment:
When inserting records with driver postgres, it's faster to use the single statement "INSERT ... RETURNING id",
rather than 2 statements -- one to get the current sequence value (and increment it), and another to do the insert.
Patch is against SQLObject-1.2.4/sqlobject/postgres/pgconnection.py.

----------------------------------------------------------------------

>Comment By: Oleg Broytman (phd)
Date: 2013-01-18 11:27

Message:
Good idea, thanks! Will test it. It's a major change (SQLObject currently
supports Postgres back to 8.1 while INSERT/RETURNING requires at least 8.2)
so I can only apply it to the trunk.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3601405&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3601405 ] Use "insert ... returning id" with postgres

[SourceForge.net <no...@so...>]

Patches item #3601405, was opened at 2013-01-18 10:39
Message generated for change (Tracker Item Submitted) made by kenlalonde
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3601405&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Ken Lalonde (kenlalonde)
Assigned to: Nobody/Anonymous (nobody)
Summary: Use "insert ... returning id" with postgres

Initial Comment:
When inserting records with driver postgres, it's faster to use the single statement "INSERT ... RETURNING id",
rather than 2 statements -- one to get the current sequence value (and increment it), and another to do the insert.
Patch is against SQLObject-1.2.4/sqlobject/postgres/pgconnection.py.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3601405&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
>Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Oleg Broytman (phd)
Date: 2012-12-08 07:00

Message:
Alas, doesn't work too. SQLite:
    cur.execute('INSERT INTO test (test.name) VALUES (?)', ('text',))
pysqlite2.dbapi2.OperationalError: near ".": syntax error

PostgreSQL:
    cur.execute('INSERT INTO test VALUES (test.id, test.value)', (1,
'test'))
psycopg2.ProgrammingError: invalid reference to FROM-clause entry for table
"tes
t"
LINE 1: INSERT INTO test VALUES (test.id, test.value)
                                 ^
HINT:  There is an entry for table "test", but it cannot be referenced from
this
 part of the query.

----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-23 10:26

Message:
Using a complex set of shell wrapper over py.test. The central part is
like the following:

for test in "$@"; do
   py.test "$test" -D "$TESTDB_URI"
done

If you are interested in all details I can send you an archive of all
shell scripts I use to maintain SQLObject. Give me your email; if you
don't want to publish it -- send it to me by a private message.

I'm going to try nose. The usage should be something like
python -c "import nose; nose.run_exit()"

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 09:48

Message:
Thanks.
How do you invoke the tests anyway?

----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-23 07:53

Message:
Interesting approach. I'll have a deeper look into into it and fix tests
if possible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 07:30

Message:
I wrote a simplistic test which covers some of the functionality.
Unfortunately, I wasn't able to invoke the testing suite, and I won't have
time to dive too deeply into the docs for a while.
The new solution should be more generally compatible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Oleg Broytman (phd)
Date: 2012-11-23 10:26

Message:
Using a complex set of shell wrapper over py.test. The central part is
like the following:

for test in "$@"; do
   py.test "$test" -D "$TESTDB_URI"
done

If you are interested in all details I can send you an archive of all
shell scripts I use to maintain SQLObject. Give me your email; if you
don't want to publish it -- send it to me by a private message.

I'm going to try nose. The usage should be something like
python -c "import nose; nose.run_exit()"

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 09:48

Message:
Thanks.
How do you invoke the tests anyway?

----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-23 07:53

Message:
Interesting approach. I'll have a deeper look into into it and fix tests
if possible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 07:30

Message:
I wrote a simplistic test which covers some of the functionality.
Unfortunately, I wasn't able to invoke the testing suite, and I won't have
time to dive too deeply into the docs for a while.
The new solution should be more generally compatible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 09:48

Message:
Thanks.
How do you invoke the tests anyway?

----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-23 07:53

Message:
Interesting approach. I'll have a deeper look into into it and fix tests
if possible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 07:30

Message:
I wrote a simplistic test which covers some of the functionality.
Unfortunately, I wasn't able to invoke the testing suite, and I won't have
time to dive too deeply into the docs for a while.
The new solution should be more generally compatible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
>Resolution: None
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Oleg Broytman (phd)
Date: 2012-11-23 07:53

Message:
Interesting approach. I'll have a deeper look into into it and fix tests
if possible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 07:30

Message:
I wrote a simplistic test which covers some of the functionality.
Unfortunately, I wasn't able to invoke the testing suite, and I won't have
time to dive too deeply into the docs for a while.
The new solution should be more generally compatible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-23 07:30

Message:
I wrote a simplistic test which covers some of the functionality.
Unfortunately, I wasn't able to invoke the testing suite, and I won't have
time to dive too deeply into the docs for a while.
The new solution should be more generally compatible.

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-21 09:24

Message:
(Removed invalid patch. Will try again shortly)

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:33

Message:
Another possibly more generally compatible solution would be to explicitly
state table_name.field_name for each.

Note however that the use of "`" is coded in a central function
(escape_keywords) and could easily be changed to get the quotes from the
db-specific abstraction. I have a few deadlines to fight and don't know the
inner workings to do it efficiently at this time. 

----------------------------------------------------------------------

Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Darren Janeczek (darren_janeczek)
Date: 2012-11-20 17:29

Message:
This patch actually causes another problem:
`table_name.field_name` is produced instead of `table_name`.`field_name`

My work around was to change the method to:

def escape_keywords(keyword):
    #Use this function to allow for the potential of field items with SQL
syntax
    if '.' in keyword:
        #The dot implies that this item contains the table name -- making
it legal
        return keyword
    return "`%s`" % (keyword)



----------------------------------------------------------------------

Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
>Resolution: Invalid
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
>Assigned to: Oleg Broytman (phd)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

>Comment By: Oleg Broytman (phd)
Date: 2012-11-20 12:43

Message:
Thanks. Alas, the patch quotes special names with backticks which AFAIK are
only valid on MySQL. Any idea if there are other backends that use
backticks to quote names?

Postgres, as well as most other databases use double quotes (") to quote
names. Can you rework the patch so it takes the quote character from the DB
Connection?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

[SQL-CVS] [ sqlobject-Patches-3588789 ] Columns with keyword names are problematic for some queries

[SourceForge.net <no...@so...>]

Patches item #3588789, was opened at 2012-11-20 12:30
Message generated for change (Tracker Item Submitted) made by darren_janeczek
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Darren Janeczek (darren_janeczek)
Assigned to: Nobody/Anonymous (nobody)
Summary: Columns with keyword names are problematic for some queries

Initial Comment:
Assume you create a SQLObject out of an existing table that happens to have a column named "group".
This crude patch will protect most queries that involve columns with names like "group" etc, by using a function that wraps the names with back quotes: `group`.
NOT THOROUGHLY TESTED.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540674&aid=3588789&group_id=74338