ID broker default behaviour, autoincrement, etc.

[Jeb Boniakowski <jb...@nntllc.com>]

There are a few things I don't understand about the way Turbine deals with
creating unique, sequential IDs.

If I compile and run the tdk out of the box (excepting the necessary changes
for MSSQL to three files, which are very very small), it builds and installs
fine, but I can't insert new records with the newapp into the RDF table.
The reason is is that the app generates a SQL command something like:

INSERT INTO RDF (RDF_ID...) (NULL...)

which would be fine if the table was created to use MSSQL's underlying
equivalent of autoincrement, IDENTITY, but it's not, as per
newapp-schema.xml, which states only to create the table, not what it's
idmethod should be, though it says the table should be autoincremented.
What I don't understand is that database.dtd says that for a table, the
default idmethod is "none".  What does this mean?  In my case, it means it
doesn't work.  Why not have it be "idbroker" which seems to be universally
portable (and a really good idea).

So, by default, my poor RDF table's RDF_ID column, which was told to be NOT
NULL because it's a PRIMARY KEY, is being asked to come up with a value for
itself, but wasn't told how.  Naturally, it raises an exception over this
treatment.  If I just tweak the schema to set the idMethod to idbroker, it
works fine, but I still don't understand if this is the way it should be.
It keeps the set up from working with the defaults.

On another note, the turbine page says to send mail if you are willing to
test new combinations.  We'd be happy to test MS SQL Server 7 with Turbine.
Just tell us where to get the scripts and what to do.

thanks,
jeb.


---------------------------------------------------------------------
To unsubscribe, e-mail: turbine-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: turbine-user-help@jakarta.apache.org