A few comments. On Tue, Jun 2, 2009 at 3:54 PM, JP Navarro <navarro@mcs.anl.gov> wrote:
Another interesting and useful characteristic of a flattened XML rendering is that it's conceptually SQL tables and relationships rendered in XML.
The various XML elements should correspond to GLUE 2 SQL tables, and the linkages between the XML elements should correspond to SQL foreign key references.
I also think it would be easier to map. I would strongly suggest to enforce the implementation of all relations in the XML in the way of Foreign Keys, as what we have done with LDAP. Appart from that, you can also choose to use a concrete tree or not, but at least you know you can navigate using a common way: Foreign Keys.
An intriguing way to pull all of these ideas together would be to develop a REST compliant URL interface for accessing GLUE 2 information, where that interface accesses information in a SQL rendering OR a flattened XML rendering of GLUE 2. That native SQL or flattened XML content can be returned to clients as CSV, XML, JSON, or other formats back to the REST client.
The XML output is what it will also be covered by the XML rendering. For other renderings, such as CSV of JSON we will have to make separated documents. Anyway, I think that with XML should be enough at the moment.
Client request -> GLUE 2 REST ---> GLUE 2 SQL or flattened XML repository Client response is CSV, XML, JSON, or other rendering of repository GLUE 2
JP
On Apr 24, 2009, at 9:21 AM, JP Navarro wrote:
Hi Balazs,
As others have described, there are several design, implementation, and support advantages to a decomposed/flattened approach. With unique IDs and links between the entities, it should be fairly straightforward to compose information into a large agreed/standard rendered document.
There are also discovery performance and functionality advantages and disadvantages to both approaches.
It seems to me like this implementation approach doesn't contradict the GLUE2 specification since it describes the same entities, attributes, and relationships, although I think you're right that it may lead us to modify the current rendering specification to include a decomposed one. An interesting questions is whether we have to have a single XML rendering in support of interoperability or if we can support several renderings? We're already willing to accommodate XML, LDAP, and SQL renderings, so why not accommodate multiple XML renderings too?
Regards,
JP On Apr 24, 2009, at 8:56 AM, Balazs Konya wrote:
hi,
David Horat wrote:
This is the same approach that has been taken for the LDAP implementation. To make this possible, IDs must be globally unique. I support this approach as it lets the implementation to be less coupled to the technology used and facilitates interaction between different systems.
we'd be rather careful going this way. in the past within nordugrid we found a common, agreed structure (at that time within nordugrid) very important.
i have doubts that flattening everything will give us better interoperability.
in any case, the issue/proposal worth to be discussed on a phonecall.
one more note: the rendering was an important part of the public comment specification. Turning everything upside down e.g. in xml may need glue to enter into another public comment period.
cheers, Balazs Konya _______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg
_______________________________________________ glue-wg mailing list glue-wg@ogf.org http://www.ogf.org/mailman/listinfo/glue-wg
-- David Horat Software Engineer – IT/GD – Grid Deployment Group CERN – European Organization for Nuclear Research » Where the web was born Address: 1211 Geneva - Switzerland, Office: 28/R-003 Phone +41 22 76 77996 Fax +41 22 76 68178 (fax to email service) Web: http://cern.ch/horat Web: http://davidhorat.com/ Profile: http://linkedin.com/in/davidhorat