On 7/18/07, Ersin Er <[EMAIL PROTECTED]> wrote:
Hi, First of all, before doing any coding, I think we must stabilize our thoughts on what we need and how we'll do it. So I suggest, before doing any CiDIT stuff in the code, we should document server initialization phase, configurations etc., our needs and solutions at cwiki.
Good idea. I started this stuff but got distracted. I got to much to juggle these days. I understand Alex's concerns about the limitations of current approach or
limitations that prevent us from doing further enhancements in the way do want but I think the solution suggested is not completely the right way. I agree with most of the comments from Chris and David and here are a few (map be duplication a little bit):
I'm not trying to pitch this as the prefect final way :). If I thought that I'd just implement it without spending time to open up a discussion on the ML to engage people. I wrote this description as a seed for discussion, then posted to the list to start all this discussion. The aim was to include the community. So this is just one possible approach that I can see. I'm hoping this brings many people to the table to work together collaboratively. Although LDAP is the right technology to store configuration information, Yes it is. the proposed solution is not the right way to do this. LDAP is not for
structured data.
I don't think you mean to say that. It is structured of courses with objectClasses and entries etc. You must mean something else perhaps. You cannot define relations between attributes and their values. That does not make it unstructured. That just makes it so you cannot "define relations between attributes and their values." (Which is my primary e-mail address? Which organizationalRole is related to
which organizationalUnit in my personal entry?) For example, to specify the order of interceptors, using attribute values like n-foo (where n is a positive integer) does not really look elegant.
Yeah I pulled that out of my ass. I'm not finished yet with the whole persistence mechanism we will use. But trust me it will not be this clumsy. It can be done elegantly as well. I should watch out with partial documentation: will try to complete on a local confluence then set up the page at apache. CiDIT should be just an interface for configuring ApacheDS. Yep. It can even provide some virtual entries. Although not CiDIT some diagnostic information can be exposed via virtual attributes. We should be able to configure ApacheDS in a few ways we want, JMX, LDAP,
XML, etc.
Yep. BTW, why don't we try to completely abstract this configuration stuff? (For
example, Spring can load it's configuration from both XML and property files.)
Yeah that's what I'm saying. Step one is to remove initialization code that does this stuff in the server so the dependencies are broken to the config beans. Many users (developers) are used to configure ApacheDS with server.xml. If
this is causing problems in embedded mode, we can also use Java based configuration option (in 2.1+) of Spring.
Still can have that. I thought I pointed this out in the document no? For the Real dynamic configuration problem, CiDIT does not really have much
to do itself. For on-the-fly configuration we need a service manager like OSGi and I think Spring-OSGİ can be an option there.
Yeah glad you brought that up. It is a great option. Our current design of interceptors do not support any listeners for
configuration changes. So it has its own problems.
Yep. The dynamic schema generation stuff is, although cool, a dark path. Aye but it can be made bright :). It's a good research topic on its own but we cannot use it for CiDIT before
it gets stable as given example, JDO or JAXB. Well there are a lot more to say but briefly I can say that we should document our current status, needs and solutions.
Aye thanks for your comments. Alex Cheers,
On 7/16/07, Alex Karasulu <[EMAIL PROTECTED]> wrote: > > Hi all, > > Here's that thread on discussing the CiDIT agenda. Let's take a look at > the following > link before beginning: > > > http://directory.apache.org/apacheds/1.5/configuration-in-dit-cidit.html > > Thoughts? Comments? > > Alex > > -- Ersin Er R.A. and Ph.D Student at the Dept. of Computer Eng. in Hacettepe University http://www.cs.hacettepe.edu.tr Committer and PMC Member of The Apache Directory Project http://directory.apache.org
