While developing some custom transforms for my talks at http://blogs.msdn.com/b/mattm/archive/2010/04/14/ssis-sessions-at-tech-ed-2010-in-new-orleans.aspx , I hit a problem where I was getting an a failed HRESULT (COM Exception) in my http://msdn.microsoft.com/en-us/library/microsoft.sqlserver.dts.pipeline.pipelinecomponent.preexecute.aspx () method on a call to http://msdn.microsoft.com/en-us/library/microsoft.sqlserver.dts.pipeline.wrapper.idtsbuffermanager100.findcolumnbylineageid.aspx ().
HRESULT 0xC020402D
Which http://msdn.microsoft.com/en-us/library/ms345164(SQL.100).aspx to DTS_E_NOBUFFERTYPEONSYNCOUTPUT.
The error message for that exception is
The %1 is a synchronous output and the buffer type cannot be retrieved for a synchronous output.
Which really wasn’t much help in this case.
The problematic line of code was #3:
- var output = ComponentMetaData.OutputCollection[0];
- var outputColumns = output.OutputColumnCollection;
- columnBufferIndex = BufferManager.FindColumnByLineageID(output.Buffer, outputColumns[0].LineageID);
I was looking for the LineageID of the only column my transform was adding, so (naturally) I thought that I should pass in the Buffer value for my output.
This is not the case.
Since in my case the Output was synchronous (i.e. its http://msdn.microsoft.com/en-us/library/microsoft.sqlserver.dts.pipeline.wrapper.idtsoutput100.synchronousinputid(v=SQL.105).aspx was mapped to the transform’s input), you can’t access it’s Buffer output type (since it’s not outputting its own buffers – it’s reusing the one from the input).
The fix was really simple – instead of the output http://msdn.microsoft.com/en-us/library/microsoft.sqlserver.dts.pipeline.wrapper.idtsoutput100.buffer(v=SQL.105).aspx , use the input buffer instead(ComponentMetadata.InputCollection[0].Buffer).
