Re: [Treebase-devel] we may have to give up on PK auto-generation by postgres

[Vladimir G. <vla...@du...>]

On Jan 28, 2010, at 2:20 PM, William Piel wrote:
>
>> Bill, do you mean the data migration was done on the DB2  
>> instance ... and then the result was ported to Postgres (via an SQL  
>> dump and restore?).
>
> Yes. (and as a corollary, a scary thing is that the migration  
> scripts have never been tested on the Pg instance)
>
>> If so, was the web app connected to the DB2 instance ever used to  
>> enter new data after the migration?
>
> Yes, in that we went through a beta-testing phase intermingled with  
> data migration. My recollection is that new records from beta- 
> testing increased ids in normal fashion (meaning i+1 from the prior  
> one -- not an order of magnitude greater).

I played with the SDSC UI as you suggested, and it indeed creates  
consecutive ids.

All this speaks in favor of the theory that TB2 Hibernate got confused  
by the DB2-->Postgres switch, which was done in the hope that  
Hibernate default behavior would remain the same.    It appears we  
must now figure out how to explicitly instruct TB2 DAO and Hibernate  
to behave in the way that best matches their behavior back in the DB2  
days.

If I am charged with this, I'll need to build some basic hands-on  
knowledge of Hibernate by doing at least one tutorial, go through some  
docs, and build a small project that mimics the TB2 setup, to  
understand what needs to be done.  This will take a few days.

Does anyone feel confident in their understanding of TB2 DAO and  
Hibernate to lead us along a faster path?

--Vladimir