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

[William P. <wil...@ya...>]

On Jan 27, 2010, at 2:42 PM, Hilmar Lapp wrote:

> So does this mean that Hibernate assumes that no-one else is touching the database while it is running.
> 
> Does this then also mean that we cannot touch the database (such as for inserting data) through any other means than the running instance of Hibernate while such an Hibernate is running? I.e., a Java program that also uses Hibernate but runs as a separate process won't do either. Is this a situation that we can live with, or does this need to be remedied?
> 
> My gut feeling is the latter, but I'd like to hear people's thoughts.

My thoughts exactly.  I wrote the below text before you beat me to this issue.  I'd really like to keep alternative interfaces open, if possible. e.g., I had imagined that I (or others) would create special interfaces, e.g. for doing housekeeping work without having to burden our (already bloated) java codebase. For example, I happen to know that there are tons of incorrect mappings between taxon_labels and taxa (which are data resources scoped to the entire database) and I was assuming that I could fix/edit these later -- or periodically update with new versions of NCBI, uBIO, etc. -- without having to add more and more miscellaneous features on the java side.  

bp

> So -- is it the case that records cannot be created by an alternative route? e.g. using perl DBI/DBD. In prior discussion about what kind of sequencing system to use, I had the impression that neither of the two choices precluded our ability to access the data via a route other than java/hibernate. This seems really disappointing to me, because I think it would be really valuable to be able to have alternative interfaces. 
> 
> Am I right in thinking that hibernate's sequencing method will make it very difficult to use alternative connections like DBI/DBD? Also, if we keep the "nextval" declaration as-is, does that preserve our ability to use DBI/DBD (even if it is deceiving)?  If the hibernate-generated ids are indeed an order of magnitude greater than the hibernate_sequence, then a new record created with nextval will collide with existing ids created by hibernate.
> 
> bp