that is not safe to say in general. this is exactly why we were working on rolling
back the hibernate string/text changes, which eventually didn't make it in the
release.
it depends on how the text hibernate mappings translate to the user's specific
database. for some databases this will work transparant. for others, they might
need to change the type of some of their database columns.
this is why i added a reference at the last minute yesterday on how to revert those
hibernate mapping changes.
also one column needs to be added to a table.
the lesson we learned is that hibernate mapping updates should be monitored much more
closely. e.g. with a continuous integration test.
this is a topic that we should talk about next meeting: we could introduce a job that
runs the test suite by starting from the previous database schema and running the
schema update script. then if someone introduces a db schema change, first, there is
a signal if this happens unexpectedly, secondly the developer is forced to also
update the DB migration scripts which means that we'll have db migration as part of
the development.
regards, tom.
Thomas Diesler wrote:
Folks,
is it save to say that 3.3.0 is backward compatible with 3.2.2?
If this is true, I can advise a component update for the SOA-P and there
would not need to be any work on the so jBPM SOA branches needed.
cheers
-thomas