jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oliver Kiessler <oliver.kiess...@gmail.com>
Subject Re: Domains and users
Date Wed, 15 Dec 2004 14:53:33 GMT
On Tue, 14 Dec 2004 19:22:25 +0100, David Nuescheler
<david.nuescheler@gmail.com> wrote:
> > > How about domains? Let's assume I have an enterprise with lots of
> > > divisions. Would I create a new workspace for every division or
> > > seperate them in different node branches like
> > > "/domains/enterprise_xy/a/content/...",
> > > "/domains/enterprise_xy/b/content/..."?
> > probably the latter...depends what you want to do
> hmmm... i would stress that it depends how
> much the divisions or domains should be aware
> of each other.
> keeping everything in one ws is obviously the
> closest you can have the different business
> units together, which makes a lot of sense if they
> should for example refer to each other by path.
> in some cases it might even make sense to
> have different repository instances for a & b
> in case they should not share anything, or
> just content that is replicated between repositories.
> of course everything between like having a repo
> or ws per domain or per enterprise may make
> sense too.

I think I am going to keep all (enterprise) divisions in one workspace
because I want to have references between divisions. I am also
thinking about keeping users in a "/global/users/..." node, so I can
manage users in one place and have references to them. for instance,
there might be a superuser who has global access to all content in the
repository but there might be users who have access to
"/domains/enterprise_xy/a/content/..." and
"/domains/enterprise_xy/b/content/..." but not
"/domains/enterprise_xy/c/content/...". So every "subdomain"
("/domains/enterprise_xy/a") has a "users node"
("/domains/enterprise_xy/a/users/...") with references to the
"/global/users/..." nodes.

Managing users (who have access to the knowledge management system) is
one thing, but I am not sure yet how to handle access to the JCR
repository itself.


View raw message