Hi all.
I am new to this forum and I have been dealing with Hibernate, through Hibernate Synchronizer Eclipse plugin, since 2 weeks.

I am developing a client server java app (no container) and I am dealing with an anagrafic-like table.

My question is: what could be the proper hibernate-mapping for a primary key on a single string dbfield (string = *char)? The string values are nicknames: so they are unique and eligible to be the primary key.

I don't actually need identity field which seems instead to be needed by Hibernate (for mapping java object to table rows).

How can I avoid this or, secondarily, how can I let identity field work in conjunction with my actual primary key string field?

Consider, for example, that "john70" is a primary key, actually a row identifier in the table.
A composite primary key, which would be the couple (id, nickname), would lead to the possibility of having two distint rows with the same nickname (!!):

1, "john70", ...
2, "john70", ...

Can anyone help me please?
Thanks in advance and regards.

Hibernate version: 3.0.5 (production)

SQL Server 2000 8.0

Hi Andrew,

You can use "composite-id" tag for composite key as follows:


Thanks,
Sanjay

for the first part...natural id as primary key, you can use 'assigned' generator.

http://www.hibernate.org/hib_docs/v3/re ... d-assigned

Hi guys, I wanna really thank you for your replies.

I will put in practice your advices as soon as I can, at worst on next weekend. (that's a personal project of my own, not my actual work :-) ).

Being sure that you deserve lots of credits ;-), I will let you know if your suggestions work good for my project.

See you soon,
Andrew

It must be better to have generated id as primary key, probably user will want to change nickname (mutable key is not a good candidate for primary key) just add "UNIQUE" constraint for nickname (it is a key too) , composite primary key doe's not make sence in this case.

baliukas wrote:



I think you are right.

Anyway I would like to clarify the aim of my java app for you to better advice me.

It's a web spider that collects user profiles from a web community (in which users must be registered).
Users aren't known in advance in my database. They instead have to be provisioned to my database by the spider. Moreover their nicknames are unique and immutable as requirement.

So the most typical task the spider has to execute is a "saveorupdate" of the profiles that it collects by surfing automatically and parsing the html source of the profile pages.

My question is: how can I implement proper update policy based on nickname equality? (Note that "save" isn't yet problematic by following your previous advice)

I have read in the "Equals and HashCode" document in the community area at http://www.hibernate.org/109.html that is goog thing to keep separate oid from actual business row identifier...

anar wrote:



Ok, it works!

Maybe it's not a very performing solution, but by forcing dynamic inserts and updates and no optimistic-lock strategy (the spider already manages concurrency of web requests and unicity of profiles being processed!) the performance is quite good :-)

Thank you very much, you definitevely deserve my credit!

Andrew