- From: Kingsley Idehen <kidehen@openlinksw.com>
- Date: Wed, 12 Aug 2009 08:52:40 -0400
- To: Christian Becker <chris@beckr.org>
- CC: Richard Cyganiak <richard@cyganiak.de>, d2rq-map-devel@lists.sourceforge.net, public-lod@w3.org
Christian Becker wrote: > Hi Kingsley, > JDBC indeed covers about 95% of the functionality we need. However, we > need database-specific code to do things like limiting rows, > determining optimal fetch sizes, or filtering out system tables that > shouldn't be used for vocabulary mapping. Christian, Depends on which JDBC Drivers you are using. Each item you list above (and some) was addressed in our ODBC Drivers in 1993; naturally, the same core functionality was re-exposed to JDBC when it emerged circa 1996-97. > As far as metadata is concerned, we do want to move towards using JDBC > to do most of the column type mapping, but we decided to postpone this > step as it will likely break current setups. Good to hear. I would like to see D2R working against our UDA family of JDBC Drivers which work consistently across: Oracle, DB2, Ingres, Informix, Sybase, MS SQL Server, MySQL, PostgreSQL, and any ODBC accessible data sources. They key here is to focus on ANSI SQL and to Of course I would also like to see D2R working against Virtuoso via its JDBC Driver, once this works, D2R can then leverage the Virtuoso Virtual DBMS Layer which sits atop ODBC/JDBC and deals with all the heuristics of required for not only connecting to an ODBC or JDBC data source, but also handling the complex task of heterogeneous federated SQL Queries (this includes Location aware Cost Optimaztion, XA, and Best Effort Distributed Union Joins). The RDF Views feature of Virtuoso leverages the Virtuoso Virtual DBMS layer re. deployment of Linked Data from 3rd party RDBMS data sources. Kingsley > > Cheers, > > Christian > > On Aug 11, 2009, at 8:19 PM, Kingsley Idehen wrote: > >> Richard Cyganiak wrote: >>> Kingsley, >>> >>> On 11 Aug 2009, at 18:45, Kingsley Idehen wrote: >>>> Why isn't D2R JDBC and/or ODBC based, in a generic sense? Both APIs >>>> provide enough Metadata oriented APIs for enabling a more RDBMS >>>> agnostic variant of D2R. >>> >>> Glad to inform you that D2RQ is JDBC based, and has been since 2004. >> Yes, but the problem is that is doesn't use JDBC metadata calls to >> ensure it simply works with any JDBC Driver. >> >> Of course we've tested it for JDBC against Virtuoso eons ago, >> something we do instinctively re. all JDBC or ODBC based apps. >> >> If JDBC is done right, you wouldn't have to make statements like: SQL >> Server support added (which make my hair stand). I am sure you know, >> JDBC Drivers have existed for SQL Server for eons etc.. >> >> To conclude, are you implying the following re. D2R: >> >> 1. JDBC support has evolved since 2004? >> 2. It now works with any JDBC Driver for any ANSI SQL compliant RDBMS. >> >> Kingsley >>> >>> Best, >>> Richard >>> >>> >>> >>>> >>>> As I am sure you can imagine, my hair stands (literally) whenever I >>>> encounter RDBMS specific client apps. :-) >>>> >>>> -- >>>> >>>> >>>> Regards, >>>> >>>> Kingsley Idehen Weblog: >>>> http://www.openlinksw.com/blog/~kidehen >>>> President & CEO OpenLink Software Web: http://www.openlinksw.com >>>> >>>> >>>> >>>> >>>> >>> >>> >> >> >> -- >> >> >> Regards, >> >> Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen >> President & CEO OpenLink Software Web: http://www.openlinksw.com >> >> >> > > > -- Regards, Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen President & CEO OpenLink Software Web: http://www.openlinksw.com
Received on Wednesday, 12 August 2009 12:53:24 UTC