[fdo-users] RE: RE: RE: Defining SpatialContext - how and when?
Maksim Sestic
max at geoinova.com
Tue Jul 10 05:54:26 EDT 2007
Gavin, works fine now :-) Thank you.
Regards,
Maksim Sestic
Maksim Sestic wrote:
>
> Gavin, thanks so much for the clarification, I gathered it myself that
> there must be some catch with MS SQL provider 3.1 itself (see my previous
> post) :-) Good thing is that spatial context I need to conform with
> doesn't really lie between 0,0 and 7000000,7000000, it's physical envelope
> is actually way smaller. I will try with actual projected CS envelope and
> precision of 0.0001 (GPS-measured one), and post the results back.
>
> With kindest regards,
> Maksim Sestic
>
>
> Gavin Cramer wrote:
>>
>> Ah, yes, this was with SQL Server. Sorry about that.
>>
>> I believe that you have come across a defect in SQL Server Provider 3.1.
>> Actually, it could be two defects at once (a spatial indexing
>> limitation, and a wrong error message). They were fixed in 3.2.x (for
>> the 2008 versions of Map and MapGuide). There is a workaround.
>>
>> One problem is that the wrong message is issued (it's off by one entry
>> if you peek at the catalogue). The correct message is:
>>
>> "Geometric value is outside of spatial index extents"
>>
>> This could come from two sources, one of which is the second defect:
>>
>> 1. You might really have some geometry value that is outside of the
>> extents.
>>
>> 2. You might have hit a problem in the FDO provider with spatial
>> indexing using large extents and small XY tolerance. The error message
>> would be either what you have seen, or a generic bad parameter
>> exception. There is a workaround.
>>
>> Looking at your extents and tolerance, you might encounter issue #2
>> eventually, even if it is not the cause of your current problem. The
>> basically, the spatial index internally has a 32-bit precision, which is
>> plenty for a real-world data in GIS with a reasonably well-chosen
>> extent. However, the internal check was incorrect, and some
>> calculations can go wrong if you specify a too-small XY tolerance. The
>> spatial index's precision is the main use for the spatial context's XY
>> tolerance in SQL Server Provider 3.1.
>>
>> The calculation of the smallest usable tolerance is simple: "largest
>> extent" / (2^31 - 1).
>>
>> Yes, that's 31, not 32, in that calculation. That's just being
>> defensive to avoid boundary cases. Anyway, your extent is 7000000, so
>> your smallest tolerance should be about 0.00326.
>>
>> Gavin
>>
>>
>>
>>
>> -----Original Message-----
>> From: fdo-users-bounces at lists.osgeo.org
>> [mailto:fdo-users-bounces at lists.osgeo.org] On Behalf Of Maksim Sestic
>> Sent: Monday, July 09, 2007 5:50 AM
>> To: fdo-users at lists.osgeo.org
>> Subject: [fdo-users] RE: RE: Defining SpatialContext - how and when?
>>
>>
>> Huh, funny thing but "Autodesk.SqlServer.3.1" provider connection
>> capabilities on SpatialContext report following:
>>
>> - SupportsCSysWKTFromCSysName: False
>> - SupportsMultipleSpatialContexts: True
>>
>> I guess I still need to set spatial context WKT by hand and this setting
>> shouldn't affect geometries that will get inserted during next stage.
>>
>> Anyway, I tried three different approaches and all of them failed,
>> raising
>> "Error
>> creating spatial context (null) (SRID=486518961), the WKT provided does
>> not
>> match the catalog." exception:
>>
>> 1) create DataStore, then insert new SpatialContext (with WKT string
>> defined
>> by hand) and associate schema class geometric properties with newly
>> created
>> CS (using SpatialContextAssociation).
>>
>> 2) create DataStore, then insert new SpatialContext (without WKT string)
>> and
>> associate schema class geometric properties with newly created CS (using
>> SpatialContextAssociation).
>>
>> 3) create DataStore, then delete "Default" spatial context and create a
>> new
>> "Default" one with different precision and envelope (but without WKT
>> string,
>> as in it's original settings). Geometric properties, when instantiated
>> within schema class, are set to "Default" CS by default so I don't need
>> to
>> manually set them.
>>
>> The other solution is to convert any MGI coordinate to WGS84 and let
>> them
>> into the datastore using "Default" CS, but this is actually not a good
>> one
>> since certain coordinate conversions are "one way only" when talking
>> about
>> precision. My datastore is supposed to keep GPS-measured coordinates
>> precise
>> to a few millimeters, while coordinate conversion roundtrip (MGI ->
>> WGS84 ->
>> MGI) involves overall imprecision measured in meters :-)
>>
>> Regards,
>> Maksim Sestic
>>
>>
>>
>> Gavin Cramer wrote:
>>>
>>> Ah, yes, the WKT...
>>>
>>> WKT only needs to be manually set (using
>>> FdoICreateSpatialContext::SetCoordinateSystemWkt() for FDO providers
>> that
>>> cannot generate WKT. This is the case for most providers (because
>> their
>>> underlying datastores don't have native coordinate system support).
>>>
>>> For those that do natively support coordinate systems (like Oracle),
>> the
>>> provider may also verify that the coordinate system name and WKT are
>> in
>>> the datastore's catalogue. It's generally just a blind string
>> comparison,
>>> so minor variations may cause a failure in WKT. To avoid this, just
>> omit
>>> setting the WKT attribute.
>>>
>>> There is an FDO capability to let you know whether WKT generation is
>>> supported: FdoIConnectionCapabilities::
>> SupportsCSysWKTFromCSysName().
>>>
>>> By the way, while your code for creating a LineString will work, the
>>> FdoIDirectPosition type is not yet a pooled object, and will result in
>> a
>>> lot of heap activity. I recommend looking at the other constructors
>> that
>>> are more efficient if you plan to use a lot of features. For example,
>>> there is one that takes an array of ordinates. There are accessors
>> that
>>> tend to be much faster as well. For example, GetItemByMembers() is
>> the
>>> fastest one that does all the ordinate reading for you, and
>> GetOrdinates()
>>> is the fastest if you want a raw ordinate array.
>>>
>>> Gavin
>>>
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: fdo-users-bounces at lists.osgeo.org on behalf of Maksim Sestic
>>> Sent: Sat 07-Jul-07 10:28 AM
>>> To: fdo-users at lists.osgeo.org
>>> Cc:
>>> Subject: [fdo-users] RE: Defining SpatialContext - how and when?
>>>
>>>
>>> Dear Gavin,
>>>
>>> Thanks for the tip. It's still bugging me. Here's what I'm doing:
>>>
>>> 1) create new DataStore named "GIS"
>>>
>>> 2) create SpatialContext named "MGI" using following parameters:
>>> - Name: MGI
>>> - Description: MGI
>>> - Coordinate System: MGI / Balkans zone 6
>>> - Coordinate System WKT: PROJCS["MGI / Balkans zone
>>>
>> 6",GEOGCS["MGI",DATUM["Militar_Geographische_Institute",SPHEROID["Bessel
>>
>>>
>> 1841",6377397.155,299.1528128]],PRIMEM["Greenwich",0.000000],UNIT["Decim
>> al
>>>
>> Degree",0.01745329251994330]],PROJECTION["Transverse_Mercator"],PARAMETE
>> R["Centr
>>>
>> al_Meridian",18],PARAMETER["Scale_Factor",0.9999],PARAMETER["False_Easti
>> ng",6500
>>>
>> 000.000000],PARAMETER["False_Northing",0.000000],UNIT["Meter",1.00000000
>> 0000]]
>>> - XY Tolerance: 0.0001
>>> - Z Tolerance: 0.0001
>>> - Extents Type: SpatialContextExtentType_Static
>>> - Extents range from 0,0 to 7000000,7000000
>>>
>>> 3) create schema "TESTSCHEMA" containing one class "TESTCLASS" with
>>> GeometryValue defined as:
>>> geomProperty.GeometryTypes = 7
>>> geomProperty.SpatialContextAssociation = "MGI"
>>>
>>> -------------
>>>
>>> Still, when I try to issue an Insert on some geometry ti throws "Error
>>> creating spatial context (null) (SRID=486518961), the WKT provided
>> does
>>> not
>>> match the catalog." again.
>>>
>>> Here's the geometry definition (sitting within specified spatial
>> context
>>> extents):
>>>
>>> Dim insertCommand As IInsert =
>>>
>> CType(m_Transaction.Connection.CreateCommand(CommandType.CommandType_Ins
>> ert),
>>> IInsert)
>>> insertCommand.SetFeatureClassName("TESTSCHEMA:TESTCLASS")
>>>
>>> Dim factory As AgfGeometryFactory = New AgfGeometryFactory()
>>> Dim pcollection As DirectPositionCollection = New
>>> DirectPositionCollection()
>>> pcollection.Add(factory.CreatePositionXY(6500001, 4950001))
>>> pcollection.Add(factory.CreatePositionXY(6534554.445, 4958445.342))
>>>
>>> Dim line As ILineString = factory.CreateLineString(pcollection)
>>> Dim geometryValue As GeometryValue = New
>>> GeometryValue(factory.GetAgf(line))
>>> insertCommand.PropertyValues.Add(New PropertyValue("geometry",
>>> geometryValue))
>>> insertCommand.Execute()
>>>
>>> ---------------------------
>>>
>>> Regards,
>>> Maksim Sestic
>>>
>>>
>>>
>>>
>>> Gavin Cramer wrote:
>>>>
>>>> Hello, Maksim. The "active" spatial context is a bit of a stale
>> idea,
>>>> and might become deprecated. Any provider that currently cares about
>> it
>>>> actually sets it internally, based on the relevant geometric
>> property's
>>>> spatial context association.
>>>>
>>>> The association can be set when creating a schema definition via
>>>> FdoGeometricPropertyDefinition:: SetSpatialContextAssociation().
>>>>
>>>> Gavin
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: fdo-users-bounces at lists.osgeo.org
>>>> [mailto:fdo-users-bounces at lists.osgeo.org] On Behalf Of Maksim Sestic
>>>> Sent: Saturday, July 07, 2007 8:50 AM
>>>> To: fdo-users at lists.osgeo.org
>>>> Subject: [fdo-users] Defining SpatialContext - how and when?
>>>>
>>>>
>>>> Autodesk Map 3D 2007 / MS SQL Server 2005
>>>>
>>>> I have problems inserting a new GeometryValue due to the "Default"
>>>> SpatialContext settings. "Default" SpatialContext has an envelope
>>>> (extents)
>>>> ranging from -2000000,-2000000 to 2000000,2000000, while my geometry
>>>> lies
>>>> way beyond default settings. Due to that fact, I keep receiving
>>>> following
>>>> error: "Error creating spatial context (null) (SRID=486518961), the
>> WKT
>>>> provided does not match the catalog."
>>>>
>>>> I tried to update Extents value on "Default" spatial context using
>>>> CreateSpatialContext command but it's not working (well, at least
>>>> nothing
>>>> changes in dbo.f_spatialcontextgroup table). Then I resorted to
>> creating
>>>> a
>>>> new SC.
>>>>
>>>> Now, I managed to create a new SpatialContext (named "MGI") with
>>>> associated
>>>> WKB string, proper extents settings, etc. Alas, I can not activate it
>>>> :-)
>>>> ActivateSpatialContext command does not affect present spatial
>> contexts
>>>> within DataStore - ISpatialContextReader always returns IsActive =
>> False
>>>> for
>>>> every SC present.
>>>>
>>>> I think I'm doing something wrong here, just can't figure out exactly
>>>> what... :-) At what moment do I need to create custom SpatialContext
>>>> (before
>>>> or after creating feature schema?), and how do I apply newly
>> established
>>>> SC
>>>> on present schema? How do I "force" my newly inserted geometric
>> values
>>>> "conformant" to present SC?
>>>>
>>>> Regards,
>>>> Maksim Sestic
>>>>
>>>>
>>>> --
>>>> View this message in context:
>>>>
>> http://www.nabble.com/Defining-SpatialContext---how-and-when--tf4040281s
>>>> 18162.html#a11478212
>>>> Sent from the fdo-users mailing list archive at Nabble.com.
>>>>
>>>> _______________________________________________
>>>> fdo-users mailing list
>>>> fdo-users at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>>>
>>>> _______________________________________________
>>>> fdo-users mailing list
>>>> fdo-users at lists.osgeo.org
>>>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>>>
>>>>
>>>
>>> --
>>> View this message in context:
>>>
>> http://www.nabble.com/Defining-SpatialContext---how-and-when--tf4040281s
>> 18162.html#a11478979
>>> Sent from the fdo-users mailing list archive at Nabble.com.
>>>
>>> _______________________________________________
>>> fdo-users mailing list
>>> fdo-users at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>>
>>>
>>>
>>> _______________________________________________
>>> fdo-users mailing list
>>> fdo-users at lists.osgeo.org
>>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>>
>>>
>>
>> --
>> View this message in context:
>> http://www.nabble.com/Defining-SpatialContext---how-and-when--tf4040281s
>> 18162.html#a11498844
>> Sent from the fdo-users mailing list archive at Nabble.com.
>>
>> _______________________________________________
>> fdo-users mailing list
>> fdo-users at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>
>> _______________________________________________
>> fdo-users mailing list
>> fdo-users at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/fdo-users
>>
>>
>
>
--
View this message in context: http://www.nabble.com/Defining-SpatialContext---how-and-when--tf4040281s18162.html#a11517926
Sent from the fdo-users mailing list archive at Nabble.com.
More information about the fdo-users
mailing list