I believe that there are some misunderstandings. There is an architecture
for a unified Directory Server. That server is made from a number of
components: the core JNDI backend, the local JNDI code, the network support,
ASN.1 encoding support, the wire-level protocol handlers, all of whom fit
together to build a unified project. LDAP (present), Kerberos (present),
DNS (pending), DHCP (pending), CosNaming (proposed) and other
directory-dependent wire-level protocols are built to the code JNDI backend,
sharing all of the network and backend plumbing, and adding just the pieces
for implementing their protocol. All of these services are represented by
directory schema on the backend, and common network plumbing on the front.
Okay, but just to mention one of my pet peeves: "refactoring" projects
into independent components within an open source project is like polishing
rocks: the result may be a bunch of really cool stones, but that will still
suck if what the user actually wants is a stone wall. Build the bricks, but
be sure their design goals are determined by the larger, user-driven project
rather than abstract notions of the perfect component. That's why I say
that it is better to distribute the whole thing as one product, at least
until the component interfaces are truly stable -- open source people know
how to cut and paste. [Just my opinion -- feel free to ignore.]
The term "sub-projects" should also be dropped in favor of "components."
Please do that -- it gave me the willies.
....Roy
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
