On Wed, Feb 11, 2009 at 6:00 AM, Emmanuel Venisse <[email protected]> wrote: > The diff between Continuum-1.2.2 and 1.3.1 generated by liquibase (1.8.1) is > the following:
Cool, can you share the command or process you used to produce this? > Maybe it can help you to understand db schema change. For a correct update, > I think we need to populate some table/column but it's a start. In a visual inspection of the schemaspy output for 1.2.2 and 1.3.1 the other day, I identified these changes: new table CONTINUUM_RELEASE_RESULT new table SCHEDULE_BUILD_QUEUES new table BUILD_QUEUE new table PROJECT_SCM_ROOT table PROJECT has new column SCM_RESULT_SCMRESULT_ID_OID table PROJECT_NOTIFIER has new column SEND_ON_SCM_FAILURE table SYSTEM_CONFIGURATION has new columns DISTRIBUTED_BUILD_ENABLED and NUMBER_OF_BUILDS_IN_PARALLEL That seems to match up with what liquibase found, plus setting up the foreign key relationships. I think I have a handle on *what* changed from 1.2.3 to 1.3.1, what I don't understand is how much of this change JDO can handle on its own. Why can't it add the NUMBER_OF_BUILDS_IN_PARALLEL column? Why does it complain that there is no default value, when I see one in the model? Another thing I wonder, when I compare the databases for the two versions... the new columns appear in the *middle* of tables. Is it okay to append new columns to the end of an existing table? That seems to be what the liquibase sql is doing. -- Wendy
