This sounds good to me. Keep in mind this will likely have to wait until
3.0 as it will probably break compatibility with existing stores.

-James

On Fri, 2004-10-15 at 13:20, Darren Hartford wrote:
> Looking to come to closure on this e-mail thread. I want to post to bugzilla the 
> enhancement request.
> 
> Before I do, I want to make sure the language and description is correct. 
> =============
> Problem:
> *Scope and URI Bindings are individual to each store, no central management of Scope 
> and Bindings.
> *Scope and bindings are saved differently for each individual store's nodestore 
> configuration.
> *No seperation of Scope/URI Bindings from the actual metadata causing problems with 
> read-only configured metadata stores.
> 
> Solution:
> Option 1: Move all Scope and URI Bindings to the root nodestore.
> Option 2: Create a new centralized ConfigStore for configuration of Scope and URI 
> Bindings with it's own storage medium configuration.
> =============
> 
> Would this sound correct James, Warwick, Oliver?
> -D
> 
> > -----Original Message-----
> > From: Warwick Burrows [mailto:[EMAIL PROTECTED]
> > Sent: Wednesday, October 13, 2004 12:05 PM
> > To: 'Slide Users Mailing List'
> > Subject: RE: read-only support (CD/DVD)
> > 
> > 
> > 
> > Yes, definitely, that information is probably in the bindings tables
> > (BINDINGS and PARENT_BINDINGS). And there are some issues 
> > with the design
> > right now that may benefit from what you're proposing. 
> > There's a problem at
> > the moment where, for jdbc stores, its assumed that the URIs for all
> > children of the jdbc store (in my case '/') are stored in the 
> > URI table, but
> > in cases where the child binding of the root is the mount 
> > point for another
> > store then these URIs aren't in the table. This may be 
> > because the URI of
> > the child store is assumed to be maintained by the child 
> > store itself. The
> > result is that the getID() call in the BD2RDBMSAdapter method 
> > finds no id
> > (returns 0) for the URI and the next call to use that id to 
> > get binding
> > information fails with an "id value out of range" type error. 
> > The workaround
> > is to skip further bind processing when the id is 0 and the 
> > stores seem to
> > come up and work with this workaround but there may be a 
> > greater design
> > issue there related to how child stores are tracked by the parent.
> > 
> > Warwick
> > 
> > 
> > 
> > > -----Original Message-----
> > > From: Darren Hartford [mailto:[EMAIL PROTECTED] 
> > > Sent: Wednesday, October 13, 2004 8:02 AM
> > > To: Slide Users Mailing List
> > > Subject: RE: read-only support (CD/DVD)
> > > 
> > > 
> > > Warwick, 
> > > With the jdbc stores, are any new tables created that might 
> > > be saving similar information as those .def.xml files? My 
> > > concern is that, as Oliver pointed out, this 
> > > 'store78.def.xml' file is created and saved *only* where the 
> > > metadata is saved for TxXMLFileDescriptor (no option to 
> > > change that). Something similar must be happening with JDBC stores.
> > > 
> > > With James recommendation of having the scope node be part of 
> > > the root store instead of the child store, the whereabouts of 
> > > the JDBC version of this information would be valuable in 
> > > determining if we can simply have this scope node information 
> > > be part of the root's 'nodestore', or if there should be a 
> > > new ScopeStore/ConfigStore that can be custom-configured.
> > > 
> > > Having this information stored only in the root node (using 
> > > whatever the root's nodestore configuration) should satisfy 
> > > read-only needs if this is the quicker solution, and I do not 
> > > forsee any drawbacks outside of minor confusion as we have 
> > > had on this thread.
> > > 
> > > -D
> > > 
> > > > -----Original Message-----
> > > > From: James Mason [mailto:[EMAIL PROTECTED]
> > > > Sent: Wednesday, October 13, 2004 1:48 AM
> > > > To: Slide Users Mailing List
> > > > Subject: Re: read-only support (CD/DVD)
> > > > 
> > > > 
> > > > Darren, I think the issue you're seeing is that the value you
> > > > enter for
> > > > the <scope> becomes part of the URI space in Slide. As Warwick and
> > > > Oliver mentioned, by defining a scope to be "store78" you've 
> > > > effectively
> > > > created a /store78 node with the contents of your store 
> > > > inside it. Since
> > > > the TxXMLFileDescriptorsStore uses a meaninfully named xml file to
> > > > represent each node, the store creates a store78.def.xml file.
> > > > 
> > > > I think earlier you were talking about insulating the store 
> > > from the 
> > > > scope it was defined on. I think that would probably be a 
> > > good idea. 
> > > > Effectively it would mean the store78 node would be part of 
> > > the root 
> > > > store instead of the child store. I'm not sure what making 
> > > that change 
> > > > would entail (maybe Oliver does).
> > > > 
> > > > For an immediate solution, if you create your store with the
> > > > same scope
> > > > you're planning on using after you burn your CD it should 
> > mask this
> > > > problem (there may be others, though).
> > > > 
> > > > -James
> > > > 
> > > > On Tue, 2004-10-12 at 22:31, Oliver Zeigermann wrote:
> > > > > Which is already there:
> > > > > 
> > > > > <parameter name="rootpath">e:/store/metadata</parameter>
> > > > > 
> > > > > By the way a description file of that fashion is 
> > created for every
> > > > > collection and content resource.
> > > > > 
> > > > > Oliver
> > > > > 
> > > > > Warwick Burrows schrieb:
> > > > > 
> > > > > > So basically the creation of this store78.def.xml file
> > > > only occurs for
> > > > > > TxXMLFileDescriptorsStore type stores as I don't see it
> > > > for a jdbc root
> > > > > > store. But I do have two TxXMLFileDescriptorsStores
> > > > configured as well and I
> > > > > > do see .XML files for those. Eg.
> > > > store2/metadata/files_2.def.xml and
> > > > > > store3/metadata/files_secondCollection.def.xml. So a
> > > > parameter to the
> > > > > > TxXMLFileDescriptorsStore telling it where to put this
> > > > description file
> > > > > > would be sensible.
> > > > > > 
> > > > > > Warwick
> > > > > > 
> > > > > > 
> > > 
> > > 
> > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > For additional commands, e-mail: [EMAIL PROTECTED]
> > > 
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> > 
> > 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to