sqlobject-cvs — CVS notification

[SQL-CVS] r2907 - SQLObject/trunk/sqlobject

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

Author: phd
Date: 2007-09-10 07:20:51 -0600 (Mon, 10 Sep 2007)
New Revision: 2907

Modified:
   SQLObject/trunk/sqlobject/col.py
Log:
No need to test for OverflowError - in Python 2.3+ int() automatically returns a long
in case of an overflow.


Modified: SQLObject/trunk/sqlobject/col.py
===================================================================
--- SQLObject/trunk/sqlobject/col.py	2007-09-10 04:15:02 UTC (rev 2906)
+++ SQLObject/trunk/sqlobject/col.py	2007-09-10 13:20:51 UTC (rev 2907)
@@ -569,10 +569,7 @@
         if isinstance(value, (int, long, sqlbuilder.SQLExpression)):
             return value
         try:
-            try:
-                return int(value)
-            except OverflowError: # for Python 2.2
-                return long(value)
+            return int(value)
         except:
             raise validators.Invalid("expected an int in the IntCol '%s', got %s %r instead" % \
                 (self.name, type(value), value), value, state)

[SQL-CVS] [ sqlobject-Bugs-1768433 ] Versioning broken in 0.9.1

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

Bugs item #1768433, was opened at 2007-08-06 14:01
Message generated for change (Comment added) made by azav
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1768433&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: SQLObject release (specify)
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: azav (azav)
Assigned to: Nobody/Anonymous (nobody)
Summary: Versioning broken in 0.9.1

Initial Comment:
The versions table uses the same constraints as the original table, so having for instance a unique column in a table will raise a DuplicateEntryError when a second version with the same name is created.

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

>Comment By: azav (azav)
Date: 2007-09-10 08:43

Message:
Logged In: YES 
user_id=1861531
Originator: YES

This only fixes part of the problem. For instance, version tables will
still be created with the foreign key constraints of the original table. 

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

Comment By: Nobody/Anonymous (nobody)
Date: 2007-09-07 20:37

Message:
Logged In: NO 

I stumbled across the same problem. As constraints are directly copied to
version tables, each version of a row needs to hold new values in all
unique columns. This may not be acceptable to all applications.

In order to fix this a minor change can be made to the file
/sqlobject/versioning/__init__.py. You merely need to add a few lines.
Heres the result from a diff between the original and my changed version:

48c48,51
<         columns[column] = defi.__class__(**defi._kw)
---
>             kwds = dict(defi._kw)
>             for kw in ["alternateID", "unique"]:
>                 if kw in kwds: del kwds[kw]
>             columns[column] = defi.__class__(**kwds)

I dont know if this is the way the sqlobject authors would do it, but it
seems to work. Hopefully it will result in a faster patch :-)

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1768433&group_id=74338

[SQL-CVS] [ sqlobject-Bugs-1768433 ] Versioning broken in 0.9.1

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

Bugs item #1768433, was opened at 2007-08-06 07:01
Message generated for change (Comment added) made by nobody
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1768433&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: SQLObject release (specify)
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: azav (azav)
Assigned to: Nobody/Anonymous (nobody)
Summary: Versioning broken in 0.9.1

Initial Comment:
The versions table uses the same constraints as the original table, so having for instance a unique column in a table will raise a DuplicateEntryError when a second version with the same name is created.

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

Comment By: Nobody/Anonymous (nobody)
Date: 2007-09-07 13:37

Message:
Logged In: NO 

I stumbled across the same problem. As constraints are directly copied to
version tables, each version of a row needs to hold new values in all
unique columns. This may not be acceptable to all applications.

In order to fix this a minor change can be made to the file
/sqlobject/versioning/__init__.py. You merely need to add a few lines.
Heres the result from a diff between the original and my changed version:

48c48,51
<         columns[column] = defi.__class__(**defi._kw)
---
>             kwds = dict(defi._kw)
>             for kw in ["alternateID", "unique"]:
>                 if kw in kwds: del kwds[kw]
>             columns[column] = defi.__class__(**kwds)

I dont know if this is the way the sqlobject authors would do it, but it
seems to work. Hopefully it will result in a faster patch :-)

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1768433&group_id=74338

[SQL-CVS] r2904 - in SQLObject/trunk/sqlobject: . tests versioning/test

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

Author: phd
Date: 2007-09-07 07:46:47 -0600 (Fri, 07 Sep 2007)
New Revision: 2904

Modified:
   SQLObject/trunk/sqlobject/col.py
   SQLObject/trunk/sqlobject/converters.py
   SQLObject/trunk/sqlobject/tests/test_auto.py
   SQLObject/trunk/sqlobject/tests/test_auto_old.py
   SQLObject/trunk/sqlobject/tests/test_datetime.py
   SQLObject/trunk/sqlobject/versioning/test/test_version.py
Log:
datetime is always available in Python 2.3+, no need to test it.


Modified: SQLObject/trunk/sqlobject/col.py
===================================================================
--- SQLObject/trunk/sqlobject/col.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/col.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -36,12 +36,8 @@
 NoDefault = sqlbuilder.NoDefault
 True, False = 1==1, 0==1
 
-try:
-    import datetime
-except ImportError: # Python 2.2
-    datetime_available = False
-else:
-    datetime_available = True
+import datetime
+datetime_available = True
 
 try:
     from mx import DateTime
@@ -66,18 +62,11 @@
         DateTimeType = type(DateTime.DateTime())
         TimeType = type(DateTime.DateTime.Time(DateTime.DateTime()))
 
-if datetime_available:
-    default_datetime_implementation = DATETIME_IMPLEMENTATION
-elif mxdatetime_available:
-    default_datetime_implementation = MXDATETIME_IMPLEMENTATION
-else:
-    default_datetime_implementation = None
+default_datetime_implementation = DATETIME_IMPLEMENTATION
 
 __all__ = ["datetime_available", "mxdatetime_available",
-    "default_datetime_implementation"
-]
-if datetime_available:
-    __all__.append("DATETIME_IMPLEMENTATION")
+        "default_datetime_implementation", "DATETIME_IMPLEMENTATION"]
+
 if mxdatetime_available:
     __all__.append("MXDATETIME_IMPLEMENTATION")
 
@@ -1031,43 +1020,42 @@
     baseClass = SOSetCol
 
 
-if datetime_available:
-    class DateTimeValidator(validators.DateValidator):
-        def to_python(self, value, state):
-            if value is None:
-                return None
-            if isinstance(value, (datetime.datetime, datetime.date, datetime.time, sqlbuilder.SQLExpression)):
-                return value
-            if mxdatetime_available:
-                if isinstance(value, DateTimeType):
-                    # convert mxDateTime instance to datetime
-                    if (self.format.find("%H") >= 0) or (self.format.find("%T")) >= 0:
-                        return datetime.datetime(value.year, value.month, value.day,
-                            value.hour, value.minute, int(value.second))
-                    else:
-                        return datetime.date(value.year, value.month, value.day)
-                elif isinstance(value, TimeType):
-                    # convert mxTime instance to time
-                    if self.format.find("%d") >= 0:
-                        return datetime.timedelta(seconds=value.seconds)
-                    else:
-                        return datetime.time(value.hour, value.minute, int(value.second))
-            try:
-                stime = time.strptime(value, self.format)
-            except:
-                raise validators.Invalid("expected a date/time string of the '%s' format in the DateTimeCol '%s', got %s %r instead" % \
-                    (self.format, self.name, type(value), value), value, state)
-            return datetime.datetime(*stime[:6])
+class DateTimeValidator(validators.DateValidator):
+    def to_python(self, value, state):
+        if value is None:
+            return None
+        if isinstance(value, (datetime.datetime, datetime.date, datetime.time, sqlbuilder.SQLExpression)):
+            return value
+        if mxdatetime_available:
+            if isinstance(value, DateTimeType):
+                # convert mxDateTime instance to datetime
+                if (self.format.find("%H") >= 0) or (self.format.find("%T")) >= 0:
+                    return datetime.datetime(value.year, value.month, value.day,
+                        value.hour, value.minute, int(value.second))
+                else:
+                    return datetime.date(value.year, value.month, value.day)
+            elif isinstance(value, TimeType):
+                # convert mxTime instance to time
+                if self.format.find("%d") >= 0:
+                    return datetime.timedelta(seconds=value.seconds)
+                else:
+                    return datetime.time(value.hour, value.minute, int(value.second))
+        try:
+            stime = time.strptime(value, self.format)
+        except:
+            raise validators.Invalid("expected a date/time string of the '%s' format in the DateTimeCol '%s', got %s %r instead" % \
+                (self.format, self.name, type(value), value), value, state)
+        return datetime.datetime(*stime[:6])
 
-        def from_python(self, value, state):
-            if value is None:
-                return None
-            if isinstance(value, (datetime.datetime, datetime.date, datetime.time, sqlbuilder.SQLExpression)):
-                return value
-            if hasattr(value, "strftime"):
-                return value.strftime(self.format)
-            raise validators.Invalid("expected a datetime in the DateTimeCol '%s', got %s %r instead" % \
-                (self.name, type(value), value), value, state)
+    def from_python(self, value, state):
+        if value is None:
+            return None
+        if isinstance(value, (datetime.datetime, datetime.date, datetime.time, sqlbuilder.SQLExpression)):
+            return value
+        if hasattr(value, "strftime"):
+            return value.strftime(self.format)
+        raise validators.Invalid("expected a datetime in the DateTimeCol '%s', got %s %r instead" % \
+            (self.name, type(value), value), value, state)
 
 if mxdatetime_available:
     class MXDateTimeValidator(validators.DateValidator):
@@ -1076,14 +1064,13 @@
                 return None
             if isinstance(value, (DateTimeType, TimeType, sqlbuilder.SQLExpression)):
                 return value
-            if datetime_available: # convert datetime instance to mxDateTime
-                if isinstance(value, datetime.datetime):
-                    return DateTime.DateTime(value.year, value.month, value.day,
-                        value.hour, value.minute, value.second)
-                elif isinstance(value, datetime.date):
-                    return DateTime.Date(value.year, value.month, value.day)
-                elif isinstance(value, datetime.time):
-                    return DateTime.Time(value.hour, value.minute, value.second)
+            if isinstance(value, datetime.datetime):
+                return DateTime.DateTime(value.year, value.month, value.day,
+                    value.hour, value.minute, value.second)
+            elif isinstance(value, datetime.date):
+                return DateTime.Date(value.year, value.month, value.day)
+            elif isinstance(value, datetime.time):
+                return DateTime.Time(value.hour, value.minute, value.second)
             try:
                 stime = time.strptime(value, self.format)
             except:
@@ -1154,19 +1141,19 @@
                        % DATETIME_IMPLEMENTATION)
     now = staticmethod(now)
 
-if datetime_available:
-    class DateValidator(DateTimeValidator):
-        def to_python(self, value, state):
-            if isinstance(value, datetime.datetime):
-                value = value.date()
-            if isinstance(value, datetime.date):
-                return value
-            value = super(DateValidator, self).to_python(value, state)
-            if isinstance(value, datetime.datetime):
-                value = value.date()
+
+class DateValidator(DateTimeValidator):
+    def to_python(self, value, state):
+        if isinstance(value, datetime.datetime):
+            value = value.date()
+        if isinstance(value, datetime.date):
             return value
+        value = super(DateValidator, self).to_python(value, state)
+        if isinstance(value, datetime.datetime):
+            value = value.date()
+        return value
 
-        from_python = to_python
+    from_python = to_python
 
 class SODateCol(SOCol):
     dateFormat = '%Y-%m-%d'
@@ -1215,21 +1202,20 @@
     baseClass = SODateCol
 
 
-if datetime_available:
-    class TimeValidator(DateTimeValidator):
-        def to_python(self, value, state):
-            if isinstance(value, datetime.time):
-                return value
-            if isinstance(value, datetime.timedelta):
-                if value.days:
-                    raise validators.Invalid(
-                        "the value for the TimeCol '%s' must has days=0, it has days=%d" %
-                            (self.name, value.days), value, state)
-                return datetime.time(*time.gmtime(value.seconds)[3:6])
-            value = super(TimeValidator, self).to_python(value, state)
-            if isinstance(value, datetime.datetime):
-                value = value.time()
+class TimeValidator(DateTimeValidator):
+    def to_python(self, value, state):
+        if isinstance(value, datetime.time):
             return value
+        if isinstance(value, datetime.timedelta):
+            if value.days:
+                raise validators.Invalid(
+                    "the value for the TimeCol '%s' must has days=0, it has days=%d" %
+                        (self.name, value.days), value, state)
+            return datetime.time(*time.gmtime(value.seconds)[3:6])
+        value = super(TimeValidator, self).to_python(value, state)
+        if isinstance(value, datetime.datetime):
+            value = value.time()
+        return value
 
 class SOTimeCol(SOCol):
     timeFormat = '%H:%M:%S'

Modified: SQLObject/trunk/sqlobject/converters.py
===================================================================
--- SQLObject/trunk/sqlobject/converters.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/converters.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -15,10 +15,7 @@
         DateTimeDeltaType = None
 
 import time
-try:
-    import datetime
-except ImportError:
-    datetime = None
+import datetime
 
 try:
     import Sybase
@@ -202,23 +199,22 @@
 
     registerConverter(time.struct_time, StructTimeConverter)
 
-if datetime:
-    def DateTimeConverter(value, db):
-        return "'%04d-%02d-%02d %02d:%02d:%02d'" % (
-            value.year, value.month, value.day,
-            value.hour, value.minute, value.second)
+def DateTimeConverter(value, db):
+    return "'%04d-%02d-%02d %02d:%02d:%02d'" % (
+        value.year, value.month, value.day,
+        value.hour, value.minute, value.second)
 
-    registerConverter(datetime.datetime, DateTimeConverter)
+registerConverter(datetime.datetime, DateTimeConverter)
 
-    def DateConverter(value, db):
-        return "'%04d-%02d-%02d'" % (value.year, value.month, value.day)
+def DateConverter(value, db):
+    return "'%04d-%02d-%02d'" % (value.year, value.month, value.day)
 
-    registerConverter(datetime.date, DateConverter)
+registerConverter(datetime.date, DateConverter)
 
-    def TimeConverter(value, db):
-        return "'%02d:%02d:%02d'" % (value.hour, value.minute, value.second)
+def TimeConverter(value, db):
+    return "'%02d:%02d:%02d'" % (value.hour, value.minute, value.second)
 
-    registerConverter(datetime.time, TimeConverter)
+registerConverter(datetime.time, TimeConverter)
 
 if Decimal:
     def DecimalConverter(value, db):

Modified: SQLObject/trunk/sqlobject/tests/test_auto.py
===================================================================
--- SQLObject/trunk/sqlobject/tests/test_auto.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/tests/test_auto.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -1,3 +1,6 @@
+from datetime import datetime
+now = datetime.now
+
 from sqlobject import *
 from sqlobject.tests.dbtest import *
 from sqlobject import classregistry
@@ -3,10 +6,4 @@
 from py.test import raises
 
-try:
-    from datetime import datetime
-    now = datetime.now
-except ImportError:
-    from mx.DateTime import now
-
 ########################################
 ## Dynamic column tests

Modified: SQLObject/trunk/sqlobject/tests/test_auto_old.py
===================================================================
--- SQLObject/trunk/sqlobject/tests/test_auto_old.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/tests/test_auto_old.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -1,11 +1,9 @@
+from datetime import datetime
+now = datetime.now
+
 from sqlobject import *
 from sqlobject.tests.dbtest import *
 from sqlobject import classregistry
-try:
-    from datetime import datetime
-    now = datetime.now
-except ImportError:
-    from mx.DateTime import now
 
 ########################################
 ## Auto class generation

Modified: SQLObject/trunk/sqlobject/tests/test_datetime.py
===================================================================
--- SQLObject/trunk/sqlobject/tests/test_datetime.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/tests/test_datetime.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -1,43 +1,42 @@
 from sqlobject import *
 from sqlobject.tests.dbtest import *
-from sqlobject import col
 
 ########################################
 ## Date/time columns
 ########################################
 
-if datetime_available:
-    col.default_datetime_implementation = DATETIME_IMPLEMENTATION
-    from datetime import datetime, date, time
+from sqlobject import col
+col.default_datetime_implementation = DATETIME_IMPLEMENTATION
+from datetime import datetime, date, time
 
-    class DateTime1(SQLObject):
-        col1 = DateTimeCol()
-        col2 = DateCol()
-        col3 = TimeCol()
+class DateTime1(SQLObject):
+    col1 = DateTimeCol()
+    col2 = DateCol()
+    col3 = TimeCol()
 
-    def test_dateTime():
-        setupClass(DateTime1)
-        _now = datetime.now()
-        dt1 = DateTime1(col1=_now, col2=_now, col3=_now.time())
+def test_dateTime():
+    setupClass(DateTime1)
+    _now = datetime.now()
+    dt1 = DateTime1(col1=_now, col2=_now, col3=_now.time())
 
-        assert isinstance(dt1.col1, datetime)
-        assert dt1.col1.year == _now.year
-        assert dt1.col1.month == _now.month
-        assert dt1.col1.day == _now.day
-        assert dt1.col1.hour == _now.hour
-        assert dt1.col1.minute == _now.minute
-        assert dt1.col1.second == int(_now.second)
+    assert isinstance(dt1.col1, datetime)
+    assert dt1.col1.year == _now.year
+    assert dt1.col1.month == _now.month
+    assert dt1.col1.day == _now.day
+    assert dt1.col1.hour == _now.hour
+    assert dt1.col1.minute == _now.minute
+    assert dt1.col1.second == int(_now.second)
 
-        assert isinstance(dt1.col2, date)
-        assert not isinstance(dt1.col2, datetime)
-        assert dt1.col2.year == _now.year
-        assert dt1.col2.month == _now.month
-        assert dt1.col2.day == _now.day
+    assert isinstance(dt1.col2, date)
+    assert not isinstance(dt1.col2, datetime)
+    assert dt1.col2.year == _now.year
+    assert dt1.col2.month == _now.month
+    assert dt1.col2.day == _now.day
 
-        assert isinstance(dt1.col3, time)
-        assert dt1.col3.hour == _now.hour
-        assert dt1.col3.minute == _now.minute
-        assert dt1.col3.second == int(_now.second)
+    assert isinstance(dt1.col3, time)
+    assert dt1.col3.hour == _now.hour
+    assert dt1.col3.minute == _now.minute
+    assert dt1.col3.second == int(_now.second)
 
 if mxdatetime_available:
     col.default_datetime_implementation = MXDATETIME_IMPLEMENTATION

Modified: SQLObject/trunk/sqlobject/versioning/test/test_version.py
===================================================================
--- SQLObject/trunk/sqlobject/versioning/test/test_version.py	2007-09-07 13:24:48 UTC (rev 2903)
+++ SQLObject/trunk/sqlobject/versioning/test/test_version.py	2007-09-07 13:46:47 UTC (rev 2904)
@@ -9,9 +9,7 @@
 from sqlobject.versioning import Versioning
 from sqlobject.tests.dbtest import *
 
-from datetime import datetime
 
-
 class MyClass(SQLObject):
     name = StringCol()
     versions = Versioning()

[SQL-CVS] r2903 - SQLObject/docs

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

Author: phd
Date: 2007-09-07 07:24:48 -0600 (Fri, 07 Sep 2007)
New Revision: 2903

Modified:
   SQLObject/docs/News.txt
   SQLObject/docs/SQLObject.txt
Log:
Going to drop support for Python 2.2.


Modified: SQLObject/docs/News.txt
===================================================================
--- SQLObject/docs/News.txt	2007-09-06 03:11:42 UTC (rev 2902)
+++ SQLObject/docs/News.txt	2007-09-07 13:24:48 UTC (rev 2903)
@@ -13,12 +13,15 @@
 Features & Interface
 --------------------
 
+* Dropped support for Python 2.2. The minimal version of Python for
+  SQLObject is 2.3 now.
+
 * SQLBuilder Select supports the rest of SelectResults options (reversed,
   distinct, joins, etc.)
 
 * SQLObject.select() (i.e., SelectResults) and DBConnection.queryForSelect()
-  use SQLBuilder Select queries; this make all SELECTs implemented via
-  single API.
+  use SQLBuilder Select queries; this make all SELECTs implemented
+  internally via a single mechanism.
 
 * SQLBuilder Joins handle SQLExpression tables (not just str/SQLObject/Alias)
   and properly sqlrepr.
@@ -42,7 +45,7 @@
 * Added ViewSQLObject.
 
 * Added sqlmeta.getColumns() to get all the columns for a class (including
-  parent classes), excluding the column 'childName' indncluding the column
+  parent classes), excluding the column 'childName' and including the column
   'id'. sqlmeta.asDict() now uses getColumns(), so there is no need to
   override it in the inheritable sqlmeta class; this makes asDict() to work
   properly on inheritable sqlobjects.

Modified: SQLObject/docs/SQLObject.txt
===================================================================
--- SQLObject/docs/SQLObject.txt	2007-09-06 03:11:42 UTC (rev 2902)
+++ SQLObject/docs/SQLObject.txt	2007-09-07 13:24:48 UTC (rev 2903)
@@ -66,7 +66,7 @@
 .. _FreeTDS: http://www.freetds.org/
 .. _ADODBAPI: http://adodbapi.sourceforge.net/
 
-Python 2.2 or higher is required.  SQLObject makes extensive use of
+Python 2.3 or higher is required.  SQLObject makes extensive use of
 new-style classes.
 
 Compared To Other Database Wrappers
@@ -283,7 +283,7 @@
     True
 
 Columns are accessed like attributes.  (This uses the ``property``
-feature of Python 2.2, so that retrieving and setting these attributes
+feature of Python, so that retrieving and setting these attributes
 executes code).  Also note that objects are unique -- there is
 generally only one ``Person`` instance of a particular id in memory at
 any one time.  If you ask for a person by a particular ID more than
@@ -1659,12 +1659,10 @@
 You can additionally pass `logger` keyword argument which should be a name
 of the logger to use. If specified and `debug` is ``True``, SQLObject will
 write debug print statements via that logger instead of printing directly
-to console. Note that this options requires logging module which is
-available starting from Python 2.3. The argument `loglevel` allows to
-choose the logging level - it can be "debug", "info", "warning", "error",
-"critical" or "exception". In case `logger` is empty or absent SQLObject
-uses prints instead of logging; `loglevel` can be "stdout" or "stderr" in
-this case; default is "stdout".
+to console. The argument `loglevel` allows to choose the logging level - it
+can be "debug", "info", "warning", "error", "critical" or "exception". In
+case `logger` is empty or absent SQLObject uses prints instead of logging;
+`loglevel` can be "stdout" or "stderr" in this case; default is "stdout".
 
 To configure logging one can do something like that::
 

[SQL-CVS] [ sqlobject-Bugs-1764739 ] Buggs in Select

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

Bugs item #1764739, was opened at 2007-07-31 19:27
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1764739&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: General
Group: None
>Status: Closed
>Resolution: Fixed
Priority: 5
Private: No
Submitted By: Johan Carlsson (johanc)
>Assigned to: Oleg Broytmann (phd)
Summary: Buggs in Select

Initial Comment:
This is for: sqlobject-0.10dev_r2829-py2.4.egg

I've found some things that doesn't work in the
sqlbuilder.Select class.

In the __sqlrepr__ method ops['limit'] isn't used correctly:

def __sqlrepr__:
    ...
    if self.ops['limit'] is not NoDefault:
        end = start + limit

Now it seem like when using slices ops['limit'] can be None (not NoDefault). 

Also as can be noted limit is never set, I supposed it to be self.ops['limit'].
Resulting in:

NameError: global name 'limit' is not defined


Select(..., limit=2)
does give (not sure why, I might be me):

OperationalError: near "<": syntax error


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

>Comment By: Oleg Broytmann (phd)
Date: 2007-09-07 17:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

I think it's been fixed in the revision 2893.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1764739&group_id=74338

[SQL-CVS] I search for the person with serious intention 26

[Marilyn P. <lx...@bl...>]

Greetings. I search for the person {the man} with serious intention. I would like to learn you better, and flight, that you love me also. I shall write more detailed information to you then in the letter. If you are interested in it, you can write to me electronic: jul...@ra...
Here you as can look my photo: http://picasaweb.google.com/CoolJyliya/PhotoJuliya> I shall be very glad to receive from you the answer. With my best regards, Yuliya.

[SQL-CVS] Big brother is watching you.

[<caw...@ms...>]

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<body>
Do you trade files online? Then they will come after you. Search the news for RIAA and see what they are doing to people like you. They can't trace you if you use Tor. Use our free program and keep your self and the internet free and safe. <a href="http://68.47.83.162/">Download Tor</a>
</body>
</html>

[SQL-CVS] Talk about a market explosion

[<dbi...@kt...>]

Investors Go Crazy For VGPM As News Hits Tuesday.

VEGA PROMOTIONAL SYSTEMS, INC
VGPM
$0.077

Pokers number one player is now the number one teacher online. Members
are already hot for the amazing new site. Learn the ins and outs of
online poker from the most famous player in the game. This week VGPM is
hosting the World Series of Poker to promote the new site. VGPM is the
next hot technology gaming group. What ever you do, don't forget to grab
VGPM first thing Wednesday.

[SQL-CVS] The Weekend trader

[<viv...@ci...>]

VGPM hits market like a storm.

VEGA PROMOTIONAL SYS, INC
VGPM
$0.07

Online multiplayer gaming is over Billion dollar industry. One online
game pulled in over 471 Million in 2006 Watch VGPM grab a lion's share
of that market with a whole new line-up of subscriber games. Read up and
get on VPGM first thing Tuesday.

[SQL-CVS] Please review this before Tuesday

[<mat...@ma...>]

VGPM begins brewing for huge returns.

Vega Promotional Sys, Inc
V G P M
$0.07

Subscription gaming becomes billion dollar industry. 471 Million From
Word of Warcraft in 2006 got investors really excited. Watch VGPM grab a
lion's share of that market with a whole new line-up of subscriber
games. Get on VGPM Tuesday morning.

[SQL-CVS] r2894 - SQLObject/docs

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

Author: phd
Date: 2007-08-30 08:33:21 -0600 (Thu, 30 Aug 2007)
New Revision: 2894

Modified:
   SQLObject/docs/News.txt
Log:
Remove 'limit' from SelectResults after setting start/end so .clone() never see limit again.


Modified: SQLObject/docs/News.txt
===================================================================
--- SQLObject/docs/News.txt	2007-08-30 14:33:02 UTC (rev 2893)
+++ SQLObject/docs/News.txt	2007-08-30 14:33:21 UTC (rev 2894)
@@ -50,6 +50,8 @@
 SQLObject 0.9.2
 ===============
 
+* A number of changes ported from `SQLObject 0.7.9`_.
+
 SQLObject 0.9.1
 ===============
 
@@ -292,6 +294,12 @@
 SQLObject 0.7.9
 ===============
 
+Bug Fixes
+---------
+
+* Remove 'limit' from SelectResults after setting start/end so .clone()
+  never see limit again.
+
 Other Changes
 -------------
 

[SQL-CVS] r2893 - in SQLObject/trunk/sqlobject: . tests

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

Author: phd
Date: 2007-08-30 08:33:02 -0600 (Thu, 30 Aug 2007)
New Revision: 2893

Modified:
   SQLObject/trunk/sqlobject/sresults.py
   SQLObject/trunk/sqlobject/tests/test_select.py
Log:
Remove 'limit' from SelectResults after setting start/end so .clone() never see limit again.


Modified: SQLObject/trunk/sqlobject/sresults.py
===================================================================
--- SQLObject/trunk/sqlobject/sresults.py	2007-08-30 14:32:38 UTC (rev 2892)
+++ SQLObject/trunk/sqlobject/sresults.py	2007-08-30 14:33:02 UTC (rev 2893)
@@ -31,7 +31,7 @@
             assert not ops.get('start', None) and not ops.get('end', None), \
                "'limit' cannot be used with 'start' or 'end'"
             ops["start"] = 0
-            ops["end"] = ops["limit"]
+            ops["end"] = ops.pop("limit")
 
         tablesDict = sqlbuilder.tablesUsedDict(self.clause, self._getConnection().dbName)
         if clauseTables:
@@ -170,13 +170,13 @@
                 if self.ops.get('end', None) is not None \
                    and self.ops['end'] < end:
                     end = self.ops['end']
-            return self.clone(limit=None, start=start, end=end)
+            return self.clone(start=start, end=end)
         else:
             if value < 0:
                 return list(iter(self))[value]
             else:
                 start = self.ops.get('start', 0) + value
-                return list(self.clone(limit=None, start=start, end=start+1))[0]
+                return list(self.clone(start=start, end=start+1))[0]
 
     def __iter__(self):
         # @@: This could be optimized, using a simpler algorithm
@@ -208,10 +208,11 @@
 
     def count(self):
         """ Counting elements of current select results """
-        assert not self.ops.get('limit'), "'limit' is meaningless with 'distinct'"
+        assert not self.ops.get('start') and not self.ops.get('end'), \
+            "start/end/limit have no meaning with 'count'"
         assert not (self.ops.get('distinct') and (self.ops.get('start')
                                                   or self.ops.get('end'))), \
-               "distinct-counting of sliced objects is not supported"
+            "distinct-counting of sliced objects is not supported"
         if self.ops.get('distinct'):
             # Column must be specified, so we are using unique ID column.
             # COUNT(DISTINCT column) is supported by MySQL and PostgreSQL,

Modified: SQLObject/trunk/sqlobject/tests/test_select.py
===================================================================
--- SQLObject/trunk/sqlobject/tests/test_select.py	2007-08-30 14:32:38 UTC (rev 2892)
+++ SQLObject/trunk/sqlobject/tests/test_select.py	2007-08-30 14:33:02 UTC (rev 2893)
@@ -77,8 +77,7 @@
 def test_05_select_limit():
     setupIter()
     assert len(list(IterTest.select(limit=2))) == 2
-    raises(AssertionError, IterTest.select(limit=2).distinct)
-    raises(AssertionError, IterTest.select(limit=2).clone, start=1)
+    raises(AssertionError, IterTest.select(limit=2).count)
 
 def test_06_like():
     setupIter()

[SQL-CVS] r2892 - in SQLObject/branches/0.9: docs sqlobject sqlobject/tests

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

Author: phd
Date: 2007-08-30 08:32:38 -0600 (Thu, 30 Aug 2007)
New Revision: 2892

Modified:
   SQLObject/branches/0.9/docs/News.txt
   SQLObject/branches/0.9/sqlobject/sresults.py
   SQLObject/branches/0.9/sqlobject/tests/test_select.py
Log:
Remove 'limit' from SelectResults after setting start/end so .clone() never see limit again.


Modified: SQLObject/branches/0.9/docs/News.txt
===================================================================
--- SQLObject/branches/0.9/docs/News.txt	2007-08-30 14:30:52 UTC (rev 2891)
+++ SQLObject/branches/0.9/docs/News.txt	2007-08-30 14:32:38 UTC (rev 2892)
@@ -10,6 +10,8 @@
 SQLObject 0.9.2
 ===============
 
+* A number of changes ported from `SQLObject 0.7.9`_.
+
 SQLObject 0.9.1
 ===============
 
@@ -258,6 +260,12 @@
 SQLObject 0.7.9
 ===============
 
+Bug Fixes
+---------
+
+* Remove 'limit' from SelectResults after setting start/end so .clone()
+  never see limit again.
+
 Other Changes
 -------------
 

Modified: SQLObject/branches/0.9/sqlobject/sresults.py
===================================================================
--- SQLObject/branches/0.9/sqlobject/sresults.py	2007-08-30 14:30:52 UTC (rev 2891)
+++ SQLObject/branches/0.9/sqlobject/sresults.py	2007-08-30 14:32:38 UTC (rev 2892)
@@ -36,6 +36,7 @@
                "'limit' cannot be used with 'start' or 'end'"
             ops["start"] = 0
             ops["end"] = ops["limit"]
+            del ops["limit"]
 
     def __repr__(self):
         return "<%s at %x>" % (self.__class__.__name__, id(self))
@@ -150,13 +151,13 @@
                 if self.ops.get('end', None) is not None \
                    and self.ops['end'] < end:
                     end = self.ops['end']
-            return self.clone(limit=None, start=start, end=end)
+            return self.clone(start=start, end=end)
         else:
             if value < 0:
                 return list(iter(self))[value]
             else:
                 start = self.ops.get('start', 0) + value
-                return list(self.clone(limit=None, start=start, end=start+1))[0]
+                return list(self.clone(start=start, end=start+1))[0]
 
     def __iter__(self):
         # @@: This could be optimized, using a simpler algorithm
@@ -183,10 +184,11 @@
 
     def count(self):
         """ Counting elements of current select results """
-        assert not self.ops.get('limit'), "'limit' is meaningless with 'distinct'"
+        assert not self.ops.get('start') and not self.ops.get('end'), \
+            "start/end/limit have no meaning with 'count'"
         assert not (self.ops.get('distinct') and (self.ops.get('start')
                                                   or self.ops.get('end'))), \
-               "distinct-counting of sliced objects is not supported"
+            "distinct-counting of sliced objects is not supported"
         if self.ops.get('distinct'):
             # Column must be specified, so we are using unique ID column.
             # COUNT(DISTINCT column) is supported by MySQL and PostgreSQL,

Modified: SQLObject/branches/0.9/sqlobject/tests/test_select.py
===================================================================
--- SQLObject/branches/0.9/sqlobject/tests/test_select.py	2007-08-30 14:30:52 UTC (rev 2891)
+++ SQLObject/branches/0.9/sqlobject/tests/test_select.py	2007-08-30 14:32:38 UTC (rev 2892)
@@ -77,8 +77,7 @@
 def test_05_select_limit():
     setupIter()
     assert len(list(IterTest.select(limit=2))) == 2
-    raises(AssertionError, IterTest.select(limit=2).distinct)
-    raises(AssertionError, IterTest.select(limit=2).clone, start=1)
+    raises(AssertionError, IterTest.select(limit=2).count)
 
 def test_06_like():
     setupIter()

[SQL-CVS] r2891 - in SQLObject/branches/0.8: docs sqlobject sqlobject/tests

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

Author: phd
Date: 2007-08-30 08:30:52 -0600 (Thu, 30 Aug 2007)
New Revision: 2891

Modified:
   SQLObject/branches/0.8/docs/News.txt
   SQLObject/branches/0.8/sqlobject/sresults.py
   SQLObject/branches/0.8/sqlobject/tests/test_select.py
Log:
Remove 'limit' from SelectResults after setting start/end so .clone() never see limit again.


Modified: SQLObject/branches/0.8/docs/News.txt
===================================================================
--- SQLObject/branches/0.8/docs/News.txt	2007-08-30 14:30:36 UTC (rev 2890)
+++ SQLObject/branches/0.8/docs/News.txt	2007-08-30 14:30:52 UTC (rev 2891)
@@ -186,6 +186,12 @@
 SQLObject 0.7.9
 ===============
 
+Bug Fixes
+---------
+
+* Remove 'limit' from SelectResults after setting start/end so .clone()
+  never see limit again.
+
 Other Changes
 -------------
 

Modified: SQLObject/branches/0.8/sqlobject/sresults.py
===================================================================
--- SQLObject/branches/0.8/sqlobject/sresults.py	2007-08-30 14:30:36 UTC (rev 2890)
+++ SQLObject/branches/0.8/sqlobject/sresults.py	2007-08-30 14:30:52 UTC (rev 2891)
@@ -36,6 +36,7 @@
                "'limit' cannot be used with 'start' or 'end'"
             ops["start"] = 0
             ops["end"] = ops["limit"]
+            del ops["limit"]
 
     def __repr__(self):
         return "<%s at %x>" % (self.__class__.__name__, id(self))
@@ -150,13 +151,13 @@
                 if self.ops.get('end', None) is not None \
                    and self.ops['end'] < end:
                     end = self.ops['end']
-            return self.clone(limit=None, start=start, end=end)
+            return self.clone(start=start, end=end)
         else:
             if value < 0:
                 return list(iter(self))[value]
             else:
                 start = self.ops.get('start', 0) + value
-                return list(self.clone(limit=None, start=start, end=start+1))[0]
+                return list(self.clone(start=start, end=start+1))[0]
 
     def __iter__(self):
         # @@: This could be optimized, using a simpler algorithm
@@ -183,10 +184,11 @@
 
     def count(self):
         """ Counting elements of current select results """
-        assert not self.ops.get('limit'), "'limit' is meaningless with 'distinct'"
+        assert not self.ops.get('start') and not self.ops.get('end'), \
+            "start/end/limit have no meaning with 'count'"
         assert not (self.ops.get('distinct') and (self.ops.get('start')
                                                   or self.ops.get('end'))), \
-               "distinct-counting of sliced objects is not supported"
+            "distinct-counting of sliced objects is not supported"
         if self.ops.get('distinct'):
             # Column must be specified, so we are using unique ID column.
             # COUNT(DISTINCT column) is supported by MySQL and PostgreSQL,

Modified: SQLObject/branches/0.8/sqlobject/tests/test_select.py
===================================================================
--- SQLObject/branches/0.8/sqlobject/tests/test_select.py	2007-08-30 14:30:36 UTC (rev 2890)
+++ SQLObject/branches/0.8/sqlobject/tests/test_select.py	2007-08-30 14:30:52 UTC (rev 2891)
@@ -77,8 +77,7 @@
 def test_05_select_limit():
     setupIter()
     assert len(list(IterTest.select(limit=2))) == 2
-    raises(AssertionError, IterTest.select(limit=2).distinct)
-    raises(AssertionError, IterTest.select(limit=2).clone, start=1)
+    raises(AssertionError, IterTest.select(limit=2).count)
 
 def test_06_like():
     setupIter()

[SQL-CVS] r2890 - in SQLObject/branches/0.7: docs sqlobject sqlobject/tests

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

Author: phd
Date: 2007-08-30 08:30:36 -0600 (Thu, 30 Aug 2007)
New Revision: 2890

Modified:
   SQLObject/branches/0.7/docs/News.txt
   SQLObject/branches/0.7/sqlobject/sresults.py
   SQLObject/branches/0.7/sqlobject/tests/test_select.py
Log:
Remove 'limit' from SelectResults after setting start/end so .clone() never see limit again.


Modified: SQLObject/branches/0.7/docs/News.txt
===================================================================
--- SQLObject/branches/0.7/docs/News.txt	2007-08-23 16:54:33 UTC (rev 2889)
+++ SQLObject/branches/0.7/docs/News.txt	2007-08-30 14:30:36 UTC (rev 2890)
@@ -10,6 +10,12 @@
 SQLObject 0.7.9
 ===============
 
+Bug Fixes
+---------
+
+* Remove 'limit' from SelectResults after setting start/end so .clone()
+  never see limit again.
+
 Other Changes
 -------------
 

Modified: SQLObject/branches/0.7/sqlobject/sresults.py
===================================================================
--- SQLObject/branches/0.7/sqlobject/sresults.py	2007-08-23 16:54:33 UTC (rev 2889)
+++ SQLObject/branches/0.7/sqlobject/sresults.py	2007-08-30 14:30:36 UTC (rev 2890)
@@ -35,6 +35,7 @@
                "'limit' cannot be used with 'start' or 'end'"
             ops["start"] = 0
             ops["end"] = ops["limit"]
+            del ops["limit"]
 
     def __repr__(self):
         return "<%s at %x>" % (self.__class__.__name__, id(self))
@@ -145,13 +146,13 @@
                 if self.ops.get('end', None) is not None \
                    and self.ops['end'] < end:
                     end = self.ops['end']
-            return self.clone(limit=None, start=start, end=end)
+            return self.clone(start=start, end=end)
         else:
             if value < 0:
                 return list(iter(self))[value]
             else:
                 start = self.ops.get('start', 0) + value
-                return list(self.clone(limit=None, start=start, end=start+1))[0]
+                return list(self.clone(start=start, end=start+1))[0]
 
     def __iter__(self):
         # @@: This could be optimized, using a simpler algorithm
@@ -178,10 +179,11 @@
 
     def count(self):
         """ Counting elements of current select results """
-        assert not self.ops.get('limit'), "'limit' is meaningless with 'distinct'"
+        assert not self.ops.get('start') and not self.ops.get('end'), \
+            "start/end/limit have no meaning with 'count'"
         assert not (self.ops.get('distinct') and (self.ops.get('start')
                                                   or self.ops.get('end'))), \
-               "distinct-counting of sliced objects is not supported"
+            "distinct-counting of sliced objects is not supported"
         if self.ops.get('distinct'):
             # Column must be specified, so we are using unique ID column.
             # COUNT(DISTINCT column) is supported by MySQL and PostgreSQL,

Modified: SQLObject/branches/0.7/sqlobject/tests/test_select.py
===================================================================
--- SQLObject/branches/0.7/sqlobject/tests/test_select.py	2007-08-23 16:54:33 UTC (rev 2889)
+++ SQLObject/branches/0.7/sqlobject/tests/test_select.py	2007-08-30 14:30:36 UTC (rev 2890)
@@ -75,8 +75,7 @@
 def test_05_select_limit():
     setupIter()
     assert len(list(IterTest.select(limit=2))) == 2
-    raises(AssertionError, IterTest.select(limit=2).distinct)
-    raises(AssertionError, IterTest.select(limit=2).clone, start=1)
+    raises(AssertionError, IterTest.select(limit=2).count)
 
 def test_06_like():
     setupIter()

[SQL-CVS] ERMX Announces Dividend Plans

[<o-f...@as...>]

Big News Hits On ERMX.

EntreMetrix Inc. (ERMX)
$0.089 UP 11.25%

Gladiator challenge is now an ERMX asset. It is one of the biggest
promoters of mixed martial arts. New dividend opportunities are expected
from this asset expansion. Read the news ASAP, and be ready to move at
open on Tuesday.

[SQL-CVS] Big News Means Big Dividends!

[<si...@gc...>]

ERMX News Hits. Plans To Expand Dividend Opportunities!

EntreMetrix Inc. (ERMX)
$0.089 UP 11.25%

Gladiator challenge is now an ERMX asset. It is one of the most prolific
promoters of live events in the fast growing sport of mixed martial
arts. One of the first steps in ERMX's over all plan of Asset expansion
to increase dividend opportunities. Check out the release and set your
buy for Tuesday morn.

[SQL-CVS] [ sqlobject-Bugs-1690105 ] UnicodeEncodeError: 'latin-1' codec can't encode characters

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

Bugs item #1690105, was opened at 2007-03-28 21:10
Message generated for change (Comment added) made by complex
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1690105&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: MySQL
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Mirinde Fekus (junas)
Assigned to: Nobody/Anonymous (nobody)
Summary: UnicodeEncodeError: 'latin-1' codec can't encode characters

Initial Comment:
I have this
http://pylonshq.com/pasties/198
after upgrading mysqldb to 1.2.2 from 1.2.1 when selecting by value that contains non-latin characters:

res = list(model.Tag.selectBy(tag = tag))

the connection string is
sqlobject.dburi = mysql://****:****@localhost/pyneohuman?sqlobject_encoding=utf-8&charset=utf8 in development.ini

so it's not latin-1...

any ideas?

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

Comment By: Viktor Ferenczi (complex)
Date: 2007-08-26 18:40

Message:
Logged In: YES 
user_id=142612
Originator: NO

My config: Python 2.5.1, SQLObject 0.9.1, MySQL 4.1.11,
MySQL-python-1.2.2
DB with InnoDB tables, utf8_general_ci everywhere.
SET NAMES utf8 issued as first SQL statement.

This bug also appears in 0.9.1. Now works fine with modified
mysql/mysqlconnection.py:

@@ -28,10 +28,6 @@
         self.user = user
         self.password = password
         self.kw = {}
-        if MySQLdb.version_info[:3] >= (1, 2, 1):
-            self.need_unicode = True
-        else:
-            self.need_unicode = False
         for key in ("unix_socket", "init_command",
                 "read_default_file", "read_default_group", "conv"):
             if key in kw:
@@ -107,10 +103,6 @@
         # done by calling ping(True) on the connection.
         for count in range(3):
             try:
-                # For MySQLdb 1.2.1 and later, we go
-                # encoding->unicode->charset (in the mysql db)
-                if self.need_unicode and not isinstance(query, unicode):
-                    query = unicode(query, self.encoding)
                 return cursor.execute(query)
             except MySQLdb.OperationalError, e:
                 if e.args[0] in (MySQLdb.constants.CR.SERVER_GONE_ERROR,
MySQLdb.constants.CR.SERVER_LOST):

and DB URI options: ?use_unicode=1&charset=utf8

It seems to me that the need_unicode stuff is somehow buggy and not
required, at least in my case.

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

Comment By: Mirinde Fekus (junas)
Date: 2007-03-29 09:35

Message:
Logged In: YES 
user_id=1304728
Originator: YES

SQLObject-0.8.1-py2.4


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

Comment By: Oleg Broytmann (phd)
Date: 2007-03-28 21:27

Message:
Logged In: YES 
user_id=4799
Originator: NO

What is the version of SQLObject?

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1690105&group_id=74338

[SQL-CVS] (no subject)

[<chr...@we...>]

Big Day Tomorrow

EntreMetrix Inc.
ERMX
$0.06

Watch ERMX Tomorrow! Keep your eyes open for big news on ERMX Thursday.
Buy low before the news hits. Get in while the price is low.

[SQL-CVS] [ sqlobject-Bugs-1665322 ] InheritableSQLObject childName column in sqlmeta.columns

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

Bugs item #1665322, was opened at 2007-02-21 18:30
Message generated for change (Settings changed) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&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: Closed
>Resolution: Wont Fix
Priority: 5
Private: No
Submitted By: Leandro Lucarella (llucax)
Assigned to: Oleg Broytmann (phd)
Summary: InheritableSQLObject childName column in sqlmeta.columns

Initial Comment:
from sqlobject import *
from sqlobject.inheritance import InheritableSQLObject

__connection__ = 'sqlite:///:memory:'

class Base(InheritableSQLObject):
        base = IntCol()
Base.createTable()

print j.Base.sqlmeta.columns
{'base': <SOIntCol base>, 'childName': <SOStringCol childName default=None>}

I think 'childName' should not be "listed" in sqlmeta.columns because it's a SQLObject "artifact" (just like 'id' column).

This complicate automatic conversion from SQLObject to other types (like TurboJson's jsonify_sqlobject() function).

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 20:12

Message:
Logged In: YES 
user_id=240225
Originator: YES

Well, I think some sort of consistency about SQLObject "artifact" columns
being in sqlmeta.columns is desirable, but since the solution is too hard
to achieve and the workarround seems reasonable, the bug could be closed. I
think Resolution should be Wont Fix, though, as the original problem, as
you said, is not really fixed.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 17:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

Initially the bug was opened with a different formulation, but if you
think it's ok to close it - let's close it.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 17:31

Message:
Logged In: YES 
user_id=240225
Originator: YES

Fair enough.

Is this bug ready to be closed?

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 17:08

Message:
Logged In: YES 
user_id=4799
Originator: NO

A new method is enough not to apply the patch to a stable branch to
prevent 0.9.1 being backward-incompatible with 0.9.2.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 17:00

Message:
Logged In: YES 
user_id=240225
Originator: YES

It adds a new method, but why do you say that changes the existing ones?
asDict() only changes for inheritable objects (for normal objects asDict()
works exactly the same) , but it was broken anyways, so I don't see how
this change could break any existing code, if existing code can't use
asDict() for inheritable objects.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 16:54

Message:
Logged In: YES 
user_id=4799
Originator: NO

The patch changes the public API - it adds a new method and changes the
existing ones - hence it can be applied only to the trunk. I also changed
the patch - mostly changed names. Applied and committed in the revisions
2885, 2886.

PS. We use py.test, not nosetest. :)

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:57

Message:
Logged In: YES 
user_id=240225
Originator: YES

Here are some untested testcases (I don't know how to use nosetest :)

They test asDict(), not getColumn() directly.
File Added: sqlobject.r2871.asDict.tests.patch

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:55

Message:
Logged In: YES 
user_id=240225
Originator: YES

This is no exactly a great patch, and doesn't fix the base problem, but it
provides a consistent way to get all the meaningfull columns for a user.

The patch adds a function getColumns() to the sqlmeta object (and the
inheritable version), that allways get all the columns for a class
(including parent classes), excluding for the 'childName' column but
including the 'id' column.

This patch makes asDict() function use getColumns(), so there is no need
to override it in the inheritable sqlmeta class. This makes asDict() to
work properly on inheritable sqlobjects, avoiding the "AttributeError:
'SomeSQLObject' object has no attribute '_SO_val_childName'" error.
File Added: sqlobject.r2871.getColumns.patch

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-28 14:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

Reopened; see bug 1665328 for details. Now waiting for a patch...

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 16:22

Message:
Logged In: YES 
user_id=240225
Originator: YES

Because it doesn't behave like a real column (1665328 is an example, a
regular column is inherited to all it's descendant, and the "None" behavior
is another example too) and it's not user data.

But I understand you don't see a great advantage on fixing it, i'll try to
work on it when I have some time if you agree something is wrong and are
interested in patches for this... (and don't want to work on a patch that
it will not be accepted).

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-23 15:35

Message:
Logged In: YES 
user_id=4799
Originator: NO

Why is it a bug? It works as advertised. It is a real column, and it
presents in .columns.

In any case it would be so hard to fix, and fixing it gives so little
advantage - I am not going to work on it. But I can test a patch if one
appears.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 15:08

Message:
Logged In: YES 
user_id=240225
Originator: YES

Ok, it should be hard to fix, but don't you agree it's a bug? Shouldn't be
emulated like the "id" column? It's not user data.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-22 22:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

SQLObject emulates "id" columns. Unlike that fictional column childName is
a real column. There is no way to remove it from .columns.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&group_id=74338

[SQL-CVS] [ sqlobject-Bugs-1665322 ] InheritableSQLObject childName column in sqlmeta.columns

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

Bugs item #1665322, was opened at 2007-02-21 12:30
Message generated for change (Comment added) made by llucax
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&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: Closed
Resolution: Fixed
Priority: 5
Private: No
Submitted By: Leandro Lucarella (llucax)
Assigned to: Oleg Broytmann (phd)
Summary: InheritableSQLObject childName column in sqlmeta.columns

Initial Comment:
from sqlobject import *
from sqlobject.inheritance import InheritableSQLObject

__connection__ = 'sqlite:///:memory:'

class Base(InheritableSQLObject):
        base = IntCol()
Base.createTable()

print j.Base.sqlmeta.columns
{'base': <SOIntCol base>, 'childName': <SOStringCol childName default=None>}

I think 'childName' should not be "listed" in sqlmeta.columns because it's a SQLObject "artifact" (just like 'id' column).

This complicate automatic conversion from SQLObject to other types (like TurboJson's jsonify_sqlobject() function).

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

>Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 13:12

Message:
Logged In: YES 
user_id=240225
Originator: YES

Well, I think some sort of consistency about SQLObject "artifact" columns
being in sqlmeta.columns is desirable, but since the solution is too hard
to achieve and the workarround seems reasonable, the bug could be closed. I
think Resolution should be Wont Fix, though, as the original problem, as
you said, is not really fixed.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 10:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

Initially the bug was opened with a different formulation, but if you
think it's ok to close it - let's close it.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 10:31

Message:
Logged In: YES 
user_id=240225
Originator: YES

Fair enough.

Is this bug ready to be closed?

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 10:08

Message:
Logged In: YES 
user_id=4799
Originator: NO

A new method is enough not to apply the patch to a stable branch to
prevent 0.9.1 being backward-incompatible with 0.9.2.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 10:00

Message:
Logged In: YES 
user_id=240225
Originator: YES

It adds a new method, but why do you say that changes the existing ones?
asDict() only changes for inheritable objects (for normal objects asDict()
works exactly the same) , but it was broken anyways, so I don't see how
this change could break any existing code, if existing code can't use
asDict() for inheritable objects.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 09:54

Message:
Logged In: YES 
user_id=4799
Originator: NO

The patch changes the public API - it adds a new method and changes the
existing ones - hence it can be applied only to the trunk. I also changed
the patch - mostly changed names. Applied and committed in the revisions
2885, 2886.

PS. We use py.test, not nosetest. :)

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-20 22:57

Message:
Logged In: YES 
user_id=240225
Originator: YES

Here are some untested testcases (I don't know how to use nosetest :)

They test asDict(), not getColumn() directly.
File Added: sqlobject.r2871.asDict.tests.patch

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-20 22:55

Message:
Logged In: YES 
user_id=240225
Originator: YES

This is no exactly a great patch, and doesn't fix the base problem, but it
provides a consistent way to get all the meaningfull columns for a user.

The patch adds a function getColumns() to the sqlmeta object (and the
inheritable version), that allways get all the columns for a class
(including parent classes), excluding for the 'childName' column but
including the 'id' column.

This patch makes asDict() function use getColumns(), so there is no need
to override it in the inheritable sqlmeta class. This makes asDict() to
work properly on inheritable sqlobjects, avoiding the "AttributeError:
'SomeSQLObject' object has no attribute '_SO_val_childName'" error.
File Added: sqlobject.r2871.getColumns.patch

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-28 08:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

Reopened; see bug 1665328 for details. Now waiting for a patch...

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 10:22

Message:
Logged In: YES 
user_id=240225
Originator: YES

Because it doesn't behave like a real column (1665328 is an example, a
regular column is inherited to all it's descendant, and the "None" behavior
is another example too) and it's not user data.

But I understand you don't see a great advantage on fixing it, i'll try to
work on it when I have some time if you agree something is wrong and are
interested in patches for this... (and don't want to work on a patch that
it will not be accepted).

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-23 09:35

Message:
Logged In: YES 
user_id=4799
Originator: NO

Why is it a bug? It works as advertised. It is a real column, and it
presents in .columns.

In any case it would be so hard to fix, and fixing it gives so little
advantage - I am not going to work on it. But I can test a patch if one
appears.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 09:08

Message:
Logged In: YES 
user_id=240225
Originator: YES

Ok, it should be hard to fix, but don't you agree it's a bug? Shouldn't be
emulated like the "id" column? It's not user data.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-22 16:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

SQLObject emulates "id" columns. Unlike that fictional column childName is
a real column. There is no way to remove it from .columns.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&group_id=74338

[SQL-CVS] [ sqlobject-Bugs-1665322 ] InheritableSQLObject childName column in sqlmeta.columns

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

Bugs item #1665322, was opened at 2007-02-21 18:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&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: Closed
>Resolution: Fixed
Priority: 5
Private: No
Submitted By: Leandro Lucarella (llucax)
Assigned to: Oleg Broytmann (phd)
Summary: InheritableSQLObject childName column in sqlmeta.columns

Initial Comment:
from sqlobject import *
from sqlobject.inheritance import InheritableSQLObject

__connection__ = 'sqlite:///:memory:'

class Base(InheritableSQLObject):
        base = IntCol()
Base.createTable()

print j.Base.sqlmeta.columns
{'base': <SOIntCol base>, 'childName': <SOStringCol childName default=None>}

I think 'childName' should not be "listed" in sqlmeta.columns because it's a SQLObject "artifact" (just like 'id' column).

This complicate automatic conversion from SQLObject to other types (like TurboJson's jsonify_sqlobject() function).

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

>Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 17:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

Initially the bug was opened with a different formulation, but if you
think it's ok to close it - let's close it.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 17:31

Message:
Logged In: YES 
user_id=240225
Originator: YES

Fair enough.

Is this bug ready to be closed?

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 17:08

Message:
Logged In: YES 
user_id=4799
Originator: NO

A new method is enough not to apply the patch to a stable branch to
prevent 0.9.1 being backward-incompatible with 0.9.2.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 17:00

Message:
Logged In: YES 
user_id=240225
Originator: YES

It adds a new method, but why do you say that changes the existing ones?
asDict() only changes for inheritable objects (for normal objects asDict()
works exactly the same) , but it was broken anyways, so I don't see how
this change could break any existing code, if existing code can't use
asDict() for inheritable objects.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 16:54

Message:
Logged In: YES 
user_id=4799
Originator: NO

The patch changes the public API - it adds a new method and changes the
existing ones - hence it can be applied only to the trunk. I also changed
the patch - mostly changed names. Applied and committed in the revisions
2885, 2886.

PS. We use py.test, not nosetest. :)

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:57

Message:
Logged In: YES 
user_id=240225
Originator: YES

Here are some untested testcases (I don't know how to use nosetest :)

They test asDict(), not getColumn() directly.
File Added: sqlobject.r2871.asDict.tests.patch

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:55

Message:
Logged In: YES 
user_id=240225
Originator: YES

This is no exactly a great patch, and doesn't fix the base problem, but it
provides a consistent way to get all the meaningfull columns for a user.

The patch adds a function getColumns() to the sqlmeta object (and the
inheritable version), that allways get all the columns for a class
(including parent classes), excluding for the 'childName' column but
including the 'id' column.

This patch makes asDict() function use getColumns(), so there is no need
to override it in the inheritable sqlmeta class. This makes asDict() to
work properly on inheritable sqlobjects, avoiding the "AttributeError:
'SomeSQLObject' object has no attribute '_SO_val_childName'" error.
File Added: sqlobject.r2871.getColumns.patch

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-28 14:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

Reopened; see bug 1665328 for details. Now waiting for a patch...

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 16:22

Message:
Logged In: YES 
user_id=240225
Originator: YES

Because it doesn't behave like a real column (1665328 is an example, a
regular column is inherited to all it's descendant, and the "None" behavior
is another example too) and it's not user data.

But I understand you don't see a great advantage on fixing it, i'll try to
work on it when I have some time if you agree something is wrong and are
interested in patches for this... (and don't want to work on a patch that
it will not be accepted).

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-23 15:35

Message:
Logged In: YES 
user_id=4799
Originator: NO

Why is it a bug? It works as advertised. It is a real column, and it
presents in .columns.

In any case it would be so hard to fix, and fixing it gives so little
advantage - I am not going to work on it. But I can test a patch if one
appears.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 15:08

Message:
Logged In: YES 
user_id=240225
Originator: YES

Ok, it should be hard to fix, but don't you agree it's a bug? Shouldn't be
emulated like the "id" column? It's not user data.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-22 22:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

SQLObject emulates "id" columns. Unlike that fictional column childName is
a real column. There is no way to remove it from .columns.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&group_id=74338

[SQL-CVS] [ sqlobject-Bugs-1665322 ] InheritableSQLObject childName column in sqlmeta.columns

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

Bugs item #1665322, was opened at 2007-02-21 12:30
Message generated for change (Comment added) made by llucax
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&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: Wont Fix
Priority: 5
Private: No
Submitted By: Leandro Lucarella (llucax)
Assigned to: Oleg Broytmann (phd)
Summary: InheritableSQLObject childName column in sqlmeta.columns

Initial Comment:
from sqlobject import *
from sqlobject.inheritance import InheritableSQLObject

__connection__ = 'sqlite:///:memory:'

class Base(InheritableSQLObject):
        base = IntCol()
Base.createTable()

print j.Base.sqlmeta.columns
{'base': <SOIntCol base>, 'childName': <SOStringCol childName default=None>}

I think 'childName' should not be "listed" in sqlmeta.columns because it's a SQLObject "artifact" (just like 'id' column).

This complicate automatic conversion from SQLObject to other types (like TurboJson's jsonify_sqlobject() function).

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

>Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 10:31

Message:
Logged In: YES 
user_id=240225
Originator: YES

Fair enough.

Is this bug ready to be closed?

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 10:08

Message:
Logged In: YES 
user_id=4799
Originator: NO

A new method is enough not to apply the patch to a stable branch to
prevent 0.9.1 being backward-incompatible with 0.9.2.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 10:00

Message:
Logged In: YES 
user_id=240225
Originator: YES

It adds a new method, but why do you say that changes the existing ones?
asDict() only changes for inheritable objects (for normal objects asDict()
works exactly the same) , but it was broken anyways, so I don't see how
this change could break any existing code, if existing code can't use
asDict() for inheritable objects.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 09:54

Message:
Logged In: YES 
user_id=4799
Originator: NO

The patch changes the public API - it adds a new method and changes the
existing ones - hence it can be applied only to the trunk. I also changed
the patch - mostly changed names. Applied and committed in the revisions
2885, 2886.

PS. We use py.test, not nosetest. :)

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-20 22:57

Message:
Logged In: YES 
user_id=240225
Originator: YES

Here are some untested testcases (I don't know how to use nosetest :)

They test asDict(), not getColumn() directly.
File Added: sqlobject.r2871.asDict.tests.patch

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-20 22:55

Message:
Logged In: YES 
user_id=240225
Originator: YES

This is no exactly a great patch, and doesn't fix the base problem, but it
provides a consistent way to get all the meaningfull columns for a user.

The patch adds a function getColumns() to the sqlmeta object (and the
inheritable version), that allways get all the columns for a class
(including parent classes), excluding for the 'childName' column but
including the 'id' column.

This patch makes asDict() function use getColumns(), so there is no need
to override it in the inheritable sqlmeta class. This makes asDict() to
work properly on inheritable sqlobjects, avoiding the "AttributeError:
'SomeSQLObject' object has no attribute '_SO_val_childName'" error.
File Added: sqlobject.r2871.getColumns.patch

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-28 08:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

Reopened; see bug 1665328 for details. Now waiting for a patch...

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 10:22

Message:
Logged In: YES 
user_id=240225
Originator: YES

Because it doesn't behave like a real column (1665328 is an example, a
regular column is inherited to all it's descendant, and the "None" behavior
is another example too) and it's not user data.

But I understand you don't see a great advantage on fixing it, i'll try to
work on it when I have some time if you agree something is wrong and are
interested in patches for this... (and don't want to work on a patch that
it will not be accepted).

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-23 09:35

Message:
Logged In: YES 
user_id=4799
Originator: NO

Why is it a bug? It works as advertised. It is a real column, and it
presents in .columns.

In any case it would be so hard to fix, and fixing it gives so little
advantage - I am not going to work on it. But I can test a patch if one
appears.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 09:08

Message:
Logged In: YES 
user_id=240225
Originator: YES

Ok, it should be hard to fix, but don't you agree it's a bug? Shouldn't be
emulated like the "id" column? It's not user data.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-22 16:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

SQLObject emulates "id" columns. Unlike that fictional column childName is
a real column. There is no way to remove it from .columns.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&group_id=74338

[SQL-CVS] [ sqlobject-Bugs-1665322 ] InheritableSQLObject childName column in sqlmeta.columns

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

Bugs item #1665322, was opened at 2007-02-21 18:30
Message generated for change (Comment added) made by phd
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&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: Wont Fix
Priority: 5
Private: No
Submitted By: Leandro Lucarella (llucax)
Assigned to: Oleg Broytmann (phd)
Summary: InheritableSQLObject childName column in sqlmeta.columns

Initial Comment:
from sqlobject import *
from sqlobject.inheritance import InheritableSQLObject

__connection__ = 'sqlite:///:memory:'

class Base(InheritableSQLObject):
        base = IntCol()
Base.createTable()

print j.Base.sqlmeta.columns
{'base': <SOIntCol base>, 'childName': <SOStringCol childName default=None>}

I think 'childName' should not be "listed" in sqlmeta.columns because it's a SQLObject "artifact" (just like 'id' column).

This complicate automatic conversion from SQLObject to other types (like TurboJson's jsonify_sqlobject() function).

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

>Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 17:08

Message:
Logged In: YES 
user_id=4799
Originator: NO

A new method is enough not to apply the patch to a stable branch to
prevent 0.9.1 being backward-incompatible with 0.9.2.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 17:00

Message:
Logged In: YES 
user_id=240225
Originator: YES

It adds a new method, but why do you say that changes the existing ones?
asDict() only changes for inheritable objects (for normal objects asDict()
works exactly the same) , but it was broken anyways, so I don't see how
this change could break any existing code, if existing code can't use
asDict() for inheritable objects.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-08-21 16:54

Message:
Logged In: YES 
user_id=4799
Originator: NO

The patch changes the public API - it adds a new method and changes the
existing ones - hence it can be applied only to the trunk. I also changed
the patch - mostly changed names. Applied and committed in the revisions
2885, 2886.

PS. We use py.test, not nosetest. :)

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:57

Message:
Logged In: YES 
user_id=240225
Originator: YES

Here are some untested testcases (I don't know how to use nosetest :)

They test asDict(), not getColumn() directly.
File Added: sqlobject.r2871.asDict.tests.patch

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-08-21 05:55

Message:
Logged In: YES 
user_id=240225
Originator: YES

This is no exactly a great patch, and doesn't fix the base problem, but it
provides a consistent way to get all the meaningfull columns for a user.

The patch adds a function getColumns() to the sqlmeta object (and the
inheritable version), that allways get all the columns for a class
(including parent classes), excluding for the 'childName' column but
including the 'id' column.

This patch makes asDict() function use getColumns(), so there is no need
to override it in the inheritable sqlmeta class. This makes asDict() to
work properly on inheritable sqlobjects, avoiding the "AttributeError:
'SomeSQLObject' object has no attribute '_SO_val_childName'" error.
File Added: sqlobject.r2871.getColumns.patch

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-28 14:15

Message:
Logged In: YES 
user_id=4799
Originator: NO

Reopened; see bug 1665328 for details. Now waiting for a patch...

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 16:22

Message:
Logged In: YES 
user_id=240225
Originator: YES

Because it doesn't behave like a real column (1665328 is an example, a
regular column is inherited to all it's descendant, and the "None" behavior
is another example too) and it's not user data.

But I understand you don't see a great advantage on fixing it, i'll try to
work on it when I have some time if you agree something is wrong and are
interested in patches for this... (and don't want to work on a patch that
it will not be accepted).

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-23 15:35

Message:
Logged In: YES 
user_id=4799
Originator: NO

Why is it a bug? It works as advertised. It is a real column, and it
presents in .columns.

In any case it would be so hard to fix, and fixing it gives so little
advantage - I am not going to work on it. But I can test a patch if one
appears.

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

Comment By: Leandro Lucarella (llucax)
Date: 2007-02-23 15:08

Message:
Logged In: YES 
user_id=240225
Originator: YES

Ok, it should be hard to fix, but don't you agree it's a bug? Shouldn't be
emulated like the "id" column? It's not user data.

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

Comment By: Oleg Broytmann (phd)
Date: 2007-02-22 22:41

Message:
Logged In: YES 
user_id=4799
Originator: NO

SQLObject emulates "id" columns. Unlike that fictional column childName is
a real column. There is no way to remove it from .columns.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1665322&group_id=74338