Re: [Geotools-devel] AutoIncrementFIDMapper question

[Michael B. <mi...@lu...>]

On Apr 29, 2006, at 11:30 PM, Jody Garnett wrote:

> Michael Brasser wrote:
>> // should the ID be generated during an insert, we need to read it  
>> back
>> // and set it into the feature
>>
>> So I was assuming that the purpose of createID was to retrieve the  
>> ID that was just generated?
> We need to tag in a module maintainer on a JDBC datastore (like  
> justin), the actual datastore API will not give you a FeatureID you  
> can trust until after you have done a commit.
>
> I should ask what your workflow is that you are wanting a FeatureID  
> before you know if the features will be accepted by the datastore?
>
> Jody

I am going by uDig's usage when a new feature is drawn using the edit  
tool. This creates and runs a CreateAndSetNewFeature command, which  
in turn creates and runs an AddFeatureCommand, which includes the line:

fid=(String) resource.addFeatures(c).iterator().next();

This line eventually uses JDBCTextFeatureWriter and  
AutoIncrementFIDMapper from the Geotools package. The returned "fid"  
is later assigned to the EditGeom we are working with.

One reason AutoIncrementFIDMapper looks strange to me is that its  
createID function seems so different from the other createID  
functions for the other mappers in the  
org.geotools.data.jdbc.fidmapper package. The main differences I see  
being:

1. AutoIncrementFIDMapper is the only mapper to return a null value  
for createID (as far as I can see).
2. AutoIncrementFIDMapper is the only mapper to return the same value  
every time it is used, regardless of the feature. In other words, if  
I were to add ten new features and were using AutoIncrementFIDMapper,  
the mapper would return id=null for all ten features. For all other  
mappers, if I were to add ten new features the mapper would return a  
unique id for each feature (using "SELECT MAX..", "new UID()", etc  
based on the specific mapper).

Regards,
Michael