
Greetings GLUE WG, Summary of the discussions held in the GLUE WG session at OGF35 on Sunday, June 17, 2012. Many thanks for the strong participation with close to 20 present in person and 2 joining over Skype. Regards, JP and Shiraz Discussion Summary 1. XML Rendering ================ We discussed the pros and cons of hierarchical, hybrid, and flat. Flat rendering Pros ------------------- * We propose that elements can only appear in one place in a valid GLUE2 document ** Ensuring that elements can't be duplicated * Makes it easier to publish subsets of information * Element can be cross-referenced * Easier to generate and publish * Easier to cherry pick desired elements to import into external schemas Flat rendering Cons ------------------- * Harder to construct complex queries ** Parent/child relationships followed using nested queries Hierarchical Pros * Easier to construct complex queries * Naturally preserves parent/child relationships Hierarchical Cons ----------------- * Very hard to "slice" big document into pieces * Duplication of information may happen under different parents * Publishing subsets and merging is hard to implement * Must still use cross-references to content published in separate documents * The parent-child relationship mandated is sometimes too strict Hybrid Pros and Cons -------------------- * Inherited from both Flat and Hierarchical * Has information duplication Discussed that a flat to hierarchical translator could be developed once for the entire community if we determine that some consumers require the hierarchical rendering to perform certain queries. Adopted/Approved ---------------- # Flat rendering # Root “glue:Entities” element required for GLUE2 documents # Main elements are also global (can be Root) to enable inclusion by other schemas # Substitution groups for main element classes to facilitate consistency and adoption by other schemas Action Items ------------ # New XML rendering document in 4 weeks # New XML schema in 4 weeks # Follow OGF process to approve rendering and schema documents LDAP Rendering ============== Extensively discussed whether to publish a DIT or not. Universal agreement that we should not mandate a DIT. We will discuss in a future teleconference the pros and cons and decide whether to publish. Note that if we publish the DIT could be an implementor experience document. Naming Enumerations and Types ============================= We plan to discuss and implement a team based process for reviewing proposed Enumeration and Types for adoption and for publishing in a shared public location. Perhaps this could be part of a GLUE profile that applies to all the renderings. JSON Rendering ============== Shiraz presented the EMI Services JSON rendering. We will discuss JSON renderings in future teleconferences.

glue-wg-bounces@ogf.org [mailto:glue-wg-bounces@ogf.org] On
Behalf Of JP Navarro said: LDAP Rendering ============== Extensively discussed whether to publish a DIT or not. Universal agreement that we should not mandate a DIT. We will discuss in a future teleconference the pros and cons and decide whether to publish. Note that if we publish the DIT could be an implementor experience document.
To add to what I said - I don't see it as an either/or, but a question of how much to specify. The previous version of the document does discuss the DIT, but in general what it gives are indicative guidelines for implementations, not rules, and it doesn't specify what I regard as implementation details like the existence of resource BDIIs. As a more general point, I don't think that any of the GLUE documentation describes an information system as a whole - i.e. it defines the format in which information should be published, but it doesn't say *what* should be published or what the information transport and aggregation methods should be. Information like that should be defined in a profile document for a given Grid, not in the schema documents themselves. Stephen -- Scanned by iCritical.
participants (2)
-
JP Navarro
-
stephen.burke@stfc.ac.uk