[fdo-internals] PATCH : PostGIS provider patch 3.4

[Bruno Scott]

Hi Chuck,

For your first point, i think i have an idea of what's going on.
Actually the provider does a describschema at first connection and caches
all the table descriptions.
It does that for speed.
I think the problem would be that the internal cache is not well updated
when you delete and recreate a table. I'm not at my office for the moment
but i will take a look at this when i will be back.
We have our own unit testing engine, but the difference with yours is we
create and populate all the tables in a pre process batch job. We lauch our
test after, so the cache is always ok in our situation.

For you second point, i haven't do any testing on parameterized query at
all. They maybe not implemented yet. I will take a look at this also.

Bruno


Chuck Wilbur wrote:
> 
> Hi Bruno,
> 
> I've grabbed your binaries and plugged them into my test application (a
> small C++ app that simply uses direct calls into FDO to manipulate a data
> store). I've run into two problems so far:
> 
> 1)
> I have some initial cleanup code that makes sure the test tables/classes
> have been cleared and deleted (I can report that defect 106 appears to be
> fixed because this works now). If the test class is not present when I
> start the testing, it gets created and I can successfully insert features.
> However, if the test class is present and has to be cleared, deleted, and
> then recreated (I stepped through the code and checked the database and
> confirmed that all these steps are happening), the subsequent insertions
> fail with an exception at FdoIInsert::Execute. Perhaps this is related to
> the insert failure in a table with an srid that I just read about.
> Unhandled exception at 0x0845fcc0 in TestSuite.exe: 0xC0000005: Access
> violation reading location 0x00000000.
> 
> 2)
> I'm creating a class named testclassdata with a field named intfield with
> integer data (and other fields, including feature id and geometry). When I
> attempt to build a parameterized query against intfield (WHERE intField >
> $1) FDO throws an exception with the messages:
> String does not represent a valid filter.
> String incorrectly formatted.
> I tracked these messages to FdoParse::ParseFilter in Parse.cpp, and they
> seem to indicate that fdo_filter_yyparse is failing, but since I'm working
> off compiled binaries I can't get much farther than that. It's not even
> clear to me from reviewing the code where fdo_filter_yyparse calls into
> the PostGIS provider (nearly identical testing code with "?" instead of
> "$1" works with Autodesk's Oracle provider, so I assume the PostGIS
> provider is at fault).
> 
> Testing code (paraphrased):
> 	FdoIConnection* m_connection; // code to set up and connect omitted
> 	FdoISelect* m_pSelectCmd;
> 	try
> 	{
> 		m_pSelectCmd = dynamic_cast<FdoISelect*>(
> 			m_connection->CreateCommand(FdoCommandType_Select) );
> 		m_pSelectCmd->SetFeatureClassName( L"testclassdata" );
> 
> 		FdoPtr<FdoParameterValueCollection> params =
> 			m_pSelectCmd->GetParameterValues();
> 		FdoLiteralValue *value = FdoInt32Value::Create(10000);
> 		if ( params )
> 		{
> 			FdoStringP paramName;
> 			paramName = FdoStringP::Format(L"$1");
> 			FdoPtr<FdoParameterValue> param = params->FindItem(paramName);
> 			if ( !param )
> 			{
> 				param = FdoParameterValue::Create(paramName);
> 				params->Add(param);
> 			}
> 			param->SetValue(value);
> 			m_pSelectCmd->SetFilter( L"intfield > $1" ); // Exception occurs here
> 		}
> 	}
> 
> 
>> -----Original Message-----
>> From: Bruno Scott [mailto:bscott at geomapgis.com]
>> Sent: Thursday, April 17, 2008 8:50 AM
>> To: fdo-internals at lists.osgeo.org
>> Subject: [fdo-internals] PATCH : PostGIS provider patch 3.4
>> 
>> 
>> 
>> This is a pretty big Patch I know.
>> Many of the tickets were related in a way or another.
>> It has became very hard to have lots a small patches
>> The next patch will be : one ticket for one patch J
>> 
>> 
>> Ticket fixed by this patch
>> 
>> #94  Generate extent for features assigned to default spatial context
>> #106 PostGIS provider cannot delete a feature class
>> #171 Fdo Postgis Autogenated identity property is mandatory
>> #178 PostGIS : Can't insert in a non-feature class
>> #232 Fdo Postgis null and not null filter does not work
>> #233 Fdo Postgis in and not in filter does not work
>> #234 Fdo Postgis currently does not support anything but lowercase
>> identifiers
>> #235 Fdo Postgis Exception with insert
>> #236 Fdo Postgis does not support non spatial classes
>> #241 Implement Support for SelectAggregates, SpatialExtents and Count
>> #310 PostGIS provider : change class naming convention without ~
>> #311 PostGIS provider : mismatch between FdoGeometricTypes and
>> FdoGeometryType
>> #312 PostGIS provider : remove boost_date_time dll dependency
>> #313 PostGIS provider : can't filter,insert or update date/datetime
>> #314 PostGIS provider : check-out/check-in crashes Autodesk Map
>> 
>> 
>> I have included binaries, it’s easier to test than recompiling
>> everythingJ
>> These binaries have been tested on Map 2009
>> 
>> Patches and binaries for Branch\3.3 and for 3.2.3 will follow
>> 
>> Bruno
>> http://www.nabble.com/file/p16744011/PostGIS_trunk_patch.zip
>> PostGIS_trunk_patch.zip 
>> http://www.nabble.com/file/p16744011/PostGIS_trunk_binaries.zip
>> PostGIS_trunk_binaries.zip 
>> -- 
>> View this message in context: 
>> http://www.nabble.com/PATCH-%3A-PostGIS-provider-patch-3.4-tp16744
> 011s18162p16744011.html
> Sent from the fdo-internals mailing list archive at Nabble.com.
> 
> 
> 
> _______________________________________________
> fdo-internals mailing list
> fdo-internals at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/fdo-internals
> 
> 

-- 
View this message in context: http://www.nabble.com/PATCH-%3A-PostGIS-provider-patch-3.4-tp16744011s18162p16824010.html
Sent from the fdo-internals mailing list archive at Nabble.com.