When we started to work on replication implementation, we needed a client API in order to communicate using the LDAP protocol with remote LDAP servers. We could have used some existing LDAP API, but as we have already defined a lot of base classes we would have had to convert back and forth, we thought it would be better to build the missing pieces and define our own complete LDAP API.
This is one of the reasons we started this sub-project.
Another reason is that we worked with Sun a few years ago to draw a new version of JNDI, an aborted effort, sadly. We also have been contacted by the OpenLDAP team as they wanted to know if we were ready to maintain jLDAP, as it needed some patches, and as they didn't have enough workforce to do so. Last, not least, OpenDS team is also interested to see if we can avoid duplication of efforts, as they are also defining an LDAP API.
All those events make it interesting to try to find a common way to define a new LDAP API, gathering the interest of the largest developer base possible.
As we have not finished to implement all the needed features, it's easy for us to try to figure which parts can be shared with other projects, and which parts must be configurable. A clear example is the underlying network layer: OpenDS will use Grizzly, while we will use MINA. This should be transparent for developers.
Last, not least, no one can claim to be better than all the other great people out there, and claim to be able to define the best possible API. We at The ASF believe that collaboration within a community is better than working alone. We are pleased to share what we have already done, ready to move forward in better directions if it allows the API to be more reliable, usable and at the end, easier to use for devlopers.