[Bioperl-l] bioperl-db: small change to create() instruction flow

[Hilmar Lapp]

Ignore this if you a) don't use bioperl-db, or b) don't care about nor 
rely on its inner workings.

OK, here goes. I made a change to the workflow in create() in the 
bioperl-db adaptors: if the adaptor looks up the object by unique key 
before attempting an insert (it does so if in caching mode) and it 
finds it, it will set the primary key of the object passed to the 
adaptor to the primary key of the object found and then return the 
passed object.

The change is that after setting the primary key, it no longer cascades 
down to storing all children.

This was a question I had punted on before. The rationale why I removed 
the cascade here is that it is inconsistent treatment: storing children 
potentially updates them, whereas the object itself is not updated if 
it is found by lookup.

Note that if the object is INSERTed, the insert operation will still 
cascade to the children.

The change gets rid of excessive updates and look-ups when loading 
terms and term relationships (terms are cached). I'm reporting this 
change here in case someone relied on it, but I doubt anyone does ...

	-hilmar
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------