[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [search-ws] [Issue] Need to support parameterized stored queryinvocation
Farrukh Najmi wrote: > > The current strawman spec is based on the model that the client > specifies a query using an abstract syntax (CQL) and the server build > the concrete query by mapping the abstract CQL to the concrete syntax > (e.g. SQL, XQuery etc.). > > Some standards such as ebXML RegRep, OGC WRS support a model where a > query is implemented in the server complete with all supported > predicates. Parts of the query are parameterized using paarameter > variables. Clients invoke the query by simply identifying the query by > its id and supplying a subset of parameters supported by the query. > The server prunes the query to remove any predicate for which > parameters were not supplied. > > In Parameterized Query there is a complete query pre-configured on the > server which is pruned down to a smaller query where query parameters > have been replaced with parameter values supplied by the client. > > Here is a brief overview of parameterized queries: > > * Standard queries are defined by some specification > o Implementations may define non-standard queries > * Server somehow implements a parametrized stored query > o The query may be implemented in any syntax (or even code). > Details are opaque to client > o The query may be as complex or simple as needed to be > o The query can easily support all kinds of joins, nested > queries etc. > * The query URL identifies a stored query by it id > * The query URL optionally provides one or more parameters supplying > name=value pairs for named parameters supported by the query > * The query MUST be pre-defined in the server and cannot be > constructed based on the query URL > * Server may provide an interface to define new queries dynamically > or allow their discovery. Open search is being considered to > specify discovery of parameterized stored queries > > > I can provide a more formal specification for Parameterized Queries > later this week. > > CQL and Parameterized queries have different strengths, weaknesses, > similarities and differences. > I can highlight these later this week as well. > > Perhaps we should consider supporting both models in a unified way in > our specification. This would meet the needs of several existing and > emerging standards. > I have just posted the specification for Parameterized query and the document is available at: <http://www.oasis-open.org/apps/org/workgroup/search-ws/download.php/25861/search-ws-param-query-1.0-draft1.pdf> Please note that: * Every attempt has been made to keep as consistent with current SRU based discussion draft, as possible * CQL queries may be supported as a canonical (standard) query within this specification * Keyword search queries (classic search engine) are supported as a canonical (standard) query within this specification * Default responseFormat is ATOM 1.0 (other are permitted) * There is an OpenSearch binding I would like to propose that as part of our standardization, we consider merging the two discussion drafts with the best of both specs in a unified spec for standardization. This would allow our spec to meet the needs of a broad cross-section of search needs in various standards and implementation. Thanks for your consideration. -- Respectfully, Farrukh
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]