Insert with a duplicate constraint

Question

Hello guys,
Version 5.0.97
Could I please clarify if it is possible to distinguish an insert from an update in BerkeleyDB JE (Collections API) ?
I can provide some code if you need it but I thought I'd ask the question first, as I could be doing something stupid.
In the BerkeleyDB-GSG (page 74), there is this statement:
DatabaseConfig.setSortedDuplicates() If true, duplicate records are allowed in the database. If this value is false, then putting a duplicate record into the database results in an error return from the put call. Note that this property can be set only at database creation time. Default is false.
I set this value as part of the setup mechanism:
private void initDBConfig(boolean readOnly) {
        dbCfg = new DatabaseConfig();
        dbCfg.setReadOnly(readOnly);                   // we want Read/Write
        dbCfg.setAllowCreate(true);                 // create if it does not exist
        dbCfg.setSortedDuplicates(false);           // do not allow duplicates in primaryDB
        dbCfg.setTemporary(false);                  // this must be a persistent database, not in-memory
        dbCfg.setDeferredWrite(false);              // No deferred writes, it must be transactional
//        dbCfg.setTransactional(true);               // explicitly make DB transactional TODO: make !readOnly ??
I can insert an item into the primaryDB but if I try to insert a second record using the same key, I find the new record overwrites the old and does not throw an error.
Have I misinterpreted this or am I missing a configuration key somewhere ?
I'm happy to provide an example if this helps although I suspect I am not setting things up correctly.
Thanks for any help.
Clive

Answer 1

Hi Clive,
Well, the BDB JE GSG documentation is a bit misleading in that particular section, because it does not detail all possible cases of doing the Database.put*() calls.
DatabaseConfig.setSortedDuplicates() configures the database to support duplicates (duplicates records having the same key) or not. Note that this database property is persistent and cannot be changed once set;  the default is false, that is, duplicates are not allowed.
If the database is configured to not support duplicates -- setSortedDuplicates(false) -- as in your case, then a Database.put() call to insert a record having a key that already exists in the database will result in overwriting the record with the same key, basically replacing the data associated with the key (an update);  it will not result in an error.  A Database.putNoOverwrite() call to insert a record having a key that already exists in the database will result in an OperationStatus.KEYEXIST error being returned, regardless of whether the database is configured to support duplicates or not (this is what you would want to attempt an insert). 
There's also Database.putNoDupData() which stores the key/data pair into the database if it  does not already appear in the database, but this method may only be called if the  database supports sorted duplicates.
Though, you mentioned you are using the Collections API.  Also, as your database is a primary database, then you've correctly configured it so that it does not allow duplicates.
In the JE Collection API, if the database is configured to not allow duplicates then the StoredMap.put() call will result in inserting a new record if the key does not already exist in the database or in updating the data if the key already exists in the database.  Note that the return value will be null if the key was not present or it will be the previous value associated with the key, if the key was present in the database.
So, if you've configured to not allow duplicates, and if you want to prevent a StoredMap.put() call from replacing/overwriting existing data if the key is already present, then you would need to first check if the key is present, by using Map.containsKey(). See the Adding Database Items section in the Java Collections Tutorial documentation.
Regards,
Andrei