Hi guys, I won't comment the whole mail, but just adding a point wich seems essential to me : we are discussing about using a container or not, with people thinking we should use Spring differently, and Alex saying this is not possible, and even that Spring won't be the best choice. Alas, nobody but Alex knows which is the exact initialization process, (ok, I must admit I know a part of it), so discussion about using a container to start the server is a dead end, as only one person knows the rationnals. What I suggest is that we describe the init process in a way anyone can have a clear understanding of how it works (I have started it months ago, and I didn't had time to go very far, but now it seems urgent to me)
Regarding CiDIT, which is related, the problem is quite the same : we don't exactly know what should be put in the DIT, and how to handle it. It would be interesting to list all the configuration which can fit into the DIT, compared to what should stay in a file (be it a Spring file). Some people are reluctant to ditch Server.xml, and I can understand their position.
From my POV, if we put everything in the DIT, then we must have a
standard way to inject and modify those informations. I think that LDIF or DSML files should be used, because it's easy and standard to inject such files in an LDAP server. Otherwise, we should also provide very simple command line scripts to get/set configuration elements, with may be a CL interface to do so (I don't know for you, but I don't want to type ldapadd with all the cryptic arguments each time I want to store a new parameter ...) We may have to discuss this before starting writing any piece of code. wdyt ? -- Regards, Cordialement, Emmanuel Lécharny www.iktek.com
