- From: John S. Erickson <john.erickson@hp.com>
- Date: Mon, 7 Apr 2003 12:54:54 -0400
- To: <www-rdf-dspace@w3.org>
Kevin asks: > So the question I am left with is: why do you think > that Simile clients should not be limited to web browsers? Well, for one thing, the Haystack client is not technically a "web browser" > I recognize the need to support distributed cache control > of the Simile metadata among the servers, but not between > server and client. Where does this need for a non-browser > client come from, and why does it not fit into one of the > server to server distribution cases... I think that we need to accomodate virtualization of the storage that the client interacts with. This storage should be assumed to be highly distributed, and will include resources that are locally stored (relative to the user). The purpose for this localization might be for performance reasons, and/or because the user's activity simply suggests that being "close" to the physical context makes the most sense. There could be reasons of control as well --- the user/client might be modifying a set of resources and thus (a) has a local copy and (b) causes a "lock" on "server" copies.
Received on Monday, 7 April 2003 12:57:49 UTC