[Date Prev][Date Next]
RE: Schema generator
- To: "OpenLdap Software List" <openldap-software@OpenLDAP.org>
- Subject: RE: Schema generator
- From: "David Parker" <email@example.com>
- Date: Fri, 3 Oct 2003 11:06:02 -0400
- Content-class: urn:content-classes:message
- Thread-index: AcOJuLyPW1yVTYtVRSWqs1G2Ut3cdQABZQpA
- Thread-topic: Schema generator
Another option for modelling/maintaining the LDAP schema is to define it first in XML. One advantage of this is that you can use an XML Schema-aware editor such as XML Spy or Oxygen to validate data when it is entered - referential integrity within the document, etc.
In our app we have an extensive LDAP schema, but we also have other components which depend upon knowing some of the relationships expressed in that schema, and it's easier to have that information expressed in a different way for those components, i.e., so they don't have to parse the LDAP schema format. We maintain the data in a big configuration XML file, and use XSL stylesheets to generate the LDAP schema file, and other configuration files that our various components need. We have the various config files generated from a single Makefile.
It doesn't completely answer the problem of having a non-technical or LDAP-familiar user maintaining the schema, of course, but it does insulate them from specific schema file syntax, and once you have the data in XML, there are various form-based approaches you can take to enable the process, for instance providing drop-down boxes for attributes syntaxes, etc.
I think XML Spy has hooks into UML generation, too. (I don't work for XML Spy - it's just what I happen to use! Oxygen is good, too, and much less expensive, though it lacks Spy's graphical capability).
> -----Original Message-----
> From: owner-openldap-software@OpenLDAP.org
> [mailto:owner-openldap-software@OpenLDAP.org]On Behalf Of Mike Jackson
> Sent: Friday, October 03, 2003 10:01 AM
> To: OpenLdap Software List
> Subject: Re: Schema generator
> ext Michael Ströder (firstname.lastname@example.org) wrote:
> > Back to original question:
> > Most problems users are facing when creating schema is not the
> > configuration syntax. It's rather a lack of understanding
> of basics like
> > LDAP syntaxes, matching, inheritance, etc.. I already thought about
> > implementing such a thing myself but I can't imagine how a
> GUI could really
> > support a dumb user creating a good schema. Before
> implementing it one
> > should have a look at the schema tools commercial vendors
> are shipping. I
> > don't use these tools though...
> I agree completely. Producing an LDAP schema is a data
> modelling job. Data
> modelling should be driven by use cases. Use cases can not be mapped
> to a wizard, except for maybe the most basic ones. A good UML tool,
> and some understanding of UML (and of course LDAP schema) will greatly
> speed along the process.
> I use (IBM) Rational Rose, but there are also free ones
> available, e.g.
> Argon UML. And the added value here is that most system
> architects will
> accept a UML model as the equivalent of "technical documentation".