hbm2dd code change to generate outputfile for deltas anyone?

PeteTh · **Joined:** Tue May 02, 2006 12:32 pm **Posts:** 8

When using Hibernate tools 3.2.0.beta8 to determine the database deltas using the update="true" option for hbm2ddl task, I am finding that the DDL is executed correctly and echoed to the screen, but is not saved in the outputfilename="schema_alter.sql".

I have had a look at the Code and it looks like currently there is no support for writing the deltas to a file.

Would any of the developers be kind enough to add this support ? it has also been requested on another thread. I think the code changes are very small :-

Hbm2DDLExporterTask

Code:

public void execute() {
      if(schemaUpdate) {
         SchemaUpdate update = new SchemaUpdate(parent.getConfiguration() );
         update.execute(scriptToConsole, exportToDatabase);         
      } 
      else {
         SchemaExport export = new SchemaExport(parent.getConfiguration() );
         if(outputfileName!=null) {
            export.setOutputFile(new File(getDestdir(),outputfileName).toString() );
         }
         if(delimiter!=null) {
            export.setDelimiter(delimiter);
         }

add a :-

update.setOutputFile(new File(getDestdir(),outputfileName).toString() );

to the first part of 'if' above and corresponding settor in SchemaUpdate, just like there is on the 'else' and SchemaExport.java , then

in the execute() of SchemaUpdate.java

Code:

String[] createSQL = configuration.generateSchemaUpdateScript( dialect, meta );
         for ( int j = 0; j < createSQL.length; j++ ) {

            final String sql = createSQL[j];
            try {
               if ( script ) {
                  System.out.println( sql );
               }
               if ( doUpdate ) {
                  log.debug( sql );
                  stmt.executeUpdate( sql );
               }
            }
            catch ( SQLException e ) {
               exceptions.add( e );
               log.error( "Unsuccessful: " + sql );
               log.error( e.getMessage() );
            }
         }

I guess we need some of the code like in the SchemaExport.java's to write formatted statements to a file writer.

Code:

String formatted = format( sql );
      if ( delimiter != null ) {
         formatted += delimiter;
      }
      if ( script ) {
         System.out.println( formatted );
      }
      log.debug( formatted );
      if ( outputFile != null ) {
         fileOutput.write( formatted + "\n" );
      }

If anyone could do this, it would be very much appreciated.

max · **Posted:** Mon Nov 20, 2006 8:27 pm

submit a patch to our jira with the changes and i will look at it.

bjorntj · **Joined:** Tue Mar 21, 2006 5:49 pm **Posts:** 20

Was this ever added to jira and/or implemented?

BTJ

max · **Posted:** Tue Dec 19, 2006 7:03 am

maybe it's in there, but can't recall the issue number.

PeteTh · **Joined:** Tue May 02, 2006 12:32 pm **Posts:** 8

I never actually raised this as a JIRA, as we went with another solution : we found that although the hbm2ddl update='true' task detected most schema changes i.e. new columns and new tables etc, it did not detect more subtle changes (from memory I think these were things like a new index from a @index annotation and maybe even nullability changes).

We needed a solution that detects all schema changes however minor, so we now get hbm2ddl task to generate a new DB schema DDL with the update='false' (this is the 'required' DB schema), then we use Navicat MySql tool which will do a database Stucture Synchronisation between the 'required' DB and our currently 'live' DB schema. The tool produces the alter DDL statements to change the 'live' schema to match the 'required' target structure. We run the alter statements as part of our build.