In a parallel with the entity stuff, the nextID() is generated by the SSFPM.
Which is ok since it works, the stateful stuff never miss a hit since the
hash is on the long variable which always hashes to the value instead of the
object. All goodness.
I am just sending this as a FYI, the ID is jboss specific and generated, we
could factor it out of the SSFPM, if it wasn't for the only fact that it
find a way to map the id to a file simply.
In the case of entity there is a clear distinction, yes the key , the
primaryKey of the database must be generated in these layers, but in the
case of the jboss fastKey (yes it's official name ;-) we should generate it
in the container layer (still debating where exactly).
In clear for the same sake of parallelism, the SSB should generate its
fastKey in the container stuff somewhere, the parallel makes more sense with
the introduction of the FastKey variable. I will keep it simple and just
put a int or long just like the SSB has today.
I will XP this and not refactor the SSB stuff for now and just go on.
"a cache miss? what is a cache miss?"
regards
marc
________________________
Marc Fleury
Chief Technology Officer
Telkel, Inc.
________________________
