On Oct 27, 2009, at 12:14 PM, Quintin Beukes wrote:
> Hey,
>
> Two questions on deploying.
>
> 1. What does the spec specify regarding deployment?
almost nothing useful, and IIUC the ee6 spec is backtracking and
removing the standardized deployment support. Basically there's not
spec concept of dependencies, and you get basic ops like install,
start, undeploy, redeploy (can't remember all the names). It's jsr88
if you want to investigate further. It also has slightly bizarre
support for plan editors that AFAIK few vendors have implemented and
no one uses (although our console's connector plan editor uses this
stuff).
>
> 2. The table generation of JPA providers. Does the spec specify
> anything regarding this?
nothing.
>
> I'm basically asking because if my table generation is set to
> drop-create then my tables are dropped/created whenever I restart the
> app or the server. Not only when I deploy/undeploy. I understand this
> will be even more so with the OSGi Geronimo, because from what I can
> gather from the mails the full deployment process will be executed
> whenever a bundle is started.
I hope not :-) and don't see that as a likely outcome. Currently
the osgi effort is working very similarly to the current 2.2 server in
that deployment generates a bundle with all the processed g metadata
such as the config.ser inside. Even with something like a rfc 66 web
container there's a good chance the first time you start the bundle
we'd construct such a config.ser and store it.
With openjpa there are several different settings, one is drop/create
whenever possible, one is update the schema to match the entities, one
is to do no DDL. Not sure about other jpa providers.
thanks
david jencks
>
> Q
