- From: Babich, Alan <ABabich@filenet.com>
- Date: Thu, 15 Jun 2000 12:28:10 -0700
- To: "'Stephan, Eric G'" <Eric.Stephan@pnl.gov>, www-webdav-dasl@w3.org
It is necessary to put this issue in context. Due to the swift pace of things today, especially related to the internet, one can not have the ultimate final product on release one, whether it be a spec. or a real product. It is only realistic to shoot for the essentials on the first release. If you are too ambitious, the market will do something else, and you will miss your product window entirely (For a spectacular failure of this type, recall the billions that were lost on the Motorola Iridium satellite network. It took too long, and so was obsolete by the time it was deployed, and had to be killed.) So, DASL and other WebDAV efforts have the rule "For the first release, cut features until it wouldn't be useful". This is not all bad. That way you can get real feedback ASAP on what the most important things are that need to be added from actual experience. That is the first reason DASL deliberately did not tackle hierarchical property values. (BTW: Hierarchical property values don't have to be stored in XML. Some DBMS's stored hierarchical properties decades before XML was even invented. So the WebDAV property model makes the distinction.) The second reason is that there was an XML query language effort in progress at the time. True, a query language is not a protocol, and a protocol is not a query language, so there is no conflict. Nevertheless, some people perceived a conflict. Furthermore, the DASL protocol might benefit from cross pollination from an accepted XML query language, so it might be beneficial to wait for resolution of that effort. A couple of proposals were floated about how to add query of hierarchical properties to DASL's basic search. They are proof that there is more than one way to add that feature to basic search -- a separate grammar is not needed. After discussion, it was decided that the right thing to do was to defer it to the second release. Remember, only the most basic of searches is specified in the DASL draft. Every DASL implementation will have to implement basic search and the rest of the non optional features in the first draft. Optional features are generally undesirable, and were minimized. DASL has escape mechanisms, so that any number of different query protocols can be added later. Furthermore, provision has been made for adding optional features to basic search. If DASL is accepted, then it is predictable that it will be extended on later releases. Alan Babich -----Original Message----- From: Stephan, Eric G [mailto:Eric.Stephan@pnl.gov] Sent: Thursday, June 15, 2000 10:41 AM To: www-webdav-dasl@w3.org Subject: Grammer support for querying property values that are XML structu res? My basic question is: Will (does) the DASL protocol define a grammer be used to query and return parts of a property value if that value is an XML structure? It seems to me that the DASL basic grammer is geared to query simple unstructured property values (e.g. strings, numbers, possibly dates), but not structured values such as an XML structure. To support this (if my impressions are correctabout the basic grammer) a separate grammer needs to be created to allow me to do this. Please let me know if I have overlooked something and my impressions are incorrect. If I am correct I'd like to hear from you on how to deal with XML structured values in queries, grammers that are being explored, leaving the problem to the client etc. Below I have included an example of the problems we hope DASL can solve for us. Thanks, Eric Stephan Sr. Research Scientist Battelle - Pacific Northwest National Laboratory P.O. Box 999, MS K1-96 Richland, WA 99352 Phone: 509-375-6977 Fax: 509-375-6631 Email: eric.stephan@pnl.gov Here is our problem using DAV today: Given the non-hierarchical view of properties on a given document/collection we are taking advantage of storing structured values in XML as shown below: <DAV:prop> <bogus:foo-model> <name>calculation</name> <version>1.0</version> <foo-author> <name>Math Ematics</name> <email>mailto:mathematics@bogus.com</email> </foo-author> </bogus:foo-model> </DAV:prop> Some problems that a XML value query grammer could help solve: 1) How do you query a property value stored in an XML structure that contains an XML element whose value is equal, not, equal, like, >, < etc.. Two examples come to mind: -Find collections/documents with the bogus:foo-model value whose value contains <name> = "calculation" AND <version> not equal to 0.0. -Determine which bogus:foo-model <version> is greater than 0.0 2) How do you return only part of the property value in the result of the query? Example: Return the foo-model name and foo-model foo-model foo-author name What I want the format my my query result property value to look like when it is returned from DAV: <bogus:foo-model> <name/> <foo-author> <name/> </foo-author> </bogus:foo-model>
Received on Thursday, 15 June 2000 15:30:22 UTC