On Nov 2, 2006, at 1:52 PM, Jeff Genender wrote:
What I meant is that I believe that the apis, as accessible from our
OEJB code won't change, but the internally generated code can/will.
This means that our code will likely still be fine and will probably
never need to change since the API's will not change. However, what
will change is any private/protected (internal) code. This means
if we
switch to a new jar, the code may very well need to be regenerated.
Tell you what. Take a half-hour and check out some of the changes
we've made to our tree so that it's easier to code against. If you
still feel the same, have the time and think it's worth it, I'm cool
with it.
Here are some of the customizations we've done:
http://fisheye.codehaus.org/changelog/openejb?cs=2728 //initial
commit
http://fisheye.codehaus.org/changelog/openejb?cs=2729
http://fisheye.codehaus.org/changelog/openejb?cs=2730
http://fisheye.codehaus.org/changelog/openejb?cs=2731
http://fisheye.codehaus.org/changelog/openejb?cs=2734
http://fisheye.codehaus.org/changelog/openejb?cs=2737
http://fisheye.codehaus.org/changelog/openejb?cs=2739
http://fisheye.codehaus.org/changelog/openejb?cs=2740
http://fisheye.codehaus.org/changelog/openejb?cs=2741
http://fisheye.codehaus.org/changelog/openejb?cs=2742
http://fisheye.codehaus.org/changelog/openejb?cs=2770
http://fisheye.codehaus.org/changelog/openejb?cs=2773
http://fisheye.codehaus.org/changelog/openejb?cs=2799
This was what I ran into, but then again, I was using pre-spec jars
(2EA3).
I ran into that too back in May when I discovered that the tree
generated from the maven plugin (i.e. the uncertified RI JAXB 2EA3)
wouldn't run with the published JAXB RI 2.0.1 certified jars; not
surprising as lots of things were changed in the spec before it went
final.
-David
David Blevins wrote:
On Nov 2, 2006, at 12:12 PM, Jeff Genender wrote:
David Blevins wrote:
Right, you and I are talking about the same thing. Any thoughts
on how
to avoid future headaces when code using the generated code doesn't
compile as we go to another 2.0.x version.
I think the code using the generated code will be ok. Its the
generated
code used internally with newer jars is where I think the
problems will
exist, since it seems to be the internal JAXB annotations that
are being
affected. But then again, who knows if the outter core APIs will
change :-(
I don't know I followed what you mean with inner and outer
annotations.
Right now our tree isn't a "generated" tree and only uses the
annotations from the JAXB 2.0 spec. Not sure how that fits in
with what
you're describing.
-David
