WORKFLOW for Changing CDM Model¶

• Run all tests in current trunk

• Have other projects (cdm-vaadin, cdmlib-apps) linking to cdmlib in your IDE (as project, not as library so renaming takes automatically place in these projects

• Create (local) branches in cdmlib, cdm-vaadin and cdmlib-apps called cdm-x.x where x.x is the model version

• Run all test in new branches

• Adapt version number and create SchemaUpdater (should be simplified)

  • create new package in cdmlib-persistence/eu.etaxonomy.cdm.database.update for the new version
  • create new SchemaUpdater in this package (compare with older SchemaUpdaters)
  • link method getPreviousUpdater to last used SchemaUpdater, accordingly link last used SchemaUpdater with the new SchemaUpdater
  • in CdmUpdater set getCurrenSchemaUpdater to new SchemaUpdater
  • in new SchemaUpdater set startSchemaVersion with current value in CdmMetaData.dbSchemaVersion
  • in new SchemaUpdater set endSchemaVersion with new value, adapt CdmMetaData.dbSchemaVersion with exact same value (should represent the expected release date)

• Change model, check for

  • hibernate annotations
  • JAXB annotations (on attribute and class level)
  • clone methods
  • other methods

• Run appimport/eu.etaxonomy.cdm.test.function.TestModelUpdate in CREATE mode for H2 to see if it works in general

• Run TestModelUpdate in CREATE mode for a new MySQL database to see how the model is translated to DB (inspect via any DBMS)

• Run existing tests:

  • Adapt changes in cdmlib-test/src/main/resources/dbscripts.001-cdm.h2.sql (should be automated but was difficult with hibernate envers)
  • Adapt changes in cdmlib-test/src/main/resources/eu.etaxonomy.cdm.database.schema.dataset.dtd (don't know if this can be automated)
  • Adapt predefined database loadings in cdmlib-test/src/main/resources/eu.etaxonomy.cdm.database. NOTE: these need to be copied to according packages in cdmlib-persistence, cdmlib-service and cdmlib-io (should be adapted that it uses always files from cdmlib-test)
  • adapt test datasets

• write new tests for all layers

• run tests for all layers

• write database update script in SchemaUpdater (see above), also test that hibernate_sequences is up-to-date after updates (still needed?)

  • test it

    • create DB with the old schema using appimport/eu.etaxonomy.cdm.test.function.TestModelUpdate in CREATE mode (also fills it with test data) and backup somewhere (for all database types, start with MySQL)
    • run TestModelUpdate in VALIDATE mode to test update script

• test brach with applications (Taxonomic Editor, Data Portals) on local databases, including update script

• test branch with applications (Taxonomic Editor, Data Portals) on developer server, including update script

• integrate into trunk

• test trunk with local database and developer server

• create new milestone, move all open tickets to new milestone, close old milestone

• roll-out Taxonomic Editor and CDM Server at the same time

• update uml model

• update project pages (java doc)

• document on "What has changed" sites