Testing a New Checkpoint Kind

In order to test a new checkpoint kind, developers will need to test recovery from checkpoints of this new kind as well as migration from one version to another one.

The checkpoint framework provides a main abstract checkpoint::Iterator interface which is a basic iterator on checkpoints. This interface is specialized in several concreted interfaces : checkpoint::Recoverer and checkpoint::Migrator. The first one is aimed at triggering recovery from checkpoints when the second one is aimed at triggering a migration from checkpoints.

Developers can easily control the maximum number of checkpoints involved in a single KIS transaction.

Generally speaking, those interfaces must not be used for standard checkpoints administration (see the section related to checkpoint administration).

A typical way for triggering a recovery from checkpoints is : 
          declare checkpoint::Recoverer recoverer;

          create recoverer;

          recoverer.start(
                i_nodeName  : "",
                i_types     : "",
                i_chunkSize : -1);
Similarly, triggering a migration will simply consist in : 
          declare checkpoint::Migrator migrator;

          create migrator;

          migrator.start(
                i_nodeName  : "",
                i_types     : "",
                i_chunkSize : -1);