Subject: Seeking status of 'Rep API' and 'JSR-170' shim from "Sakai Repository API Design"
Newsgroups: gmane.comp.cms.sakai.devel
Date: 2006-04-26 13:58:56 GMT (2 years, 10 weeks, 4 days, 10 hours and 6 minutes ago)
I'm still working my way around "Sakai-land" (if you will) seeking the best way to integrate a Fedora-based content repository with Sakai. I've had great conversations with Beth Kirschner from UMich and Ben Brophy from MIT about their efforts. Our desires differ from those projects, though, in that we seek to make Fedora /the/ content repository for Sakai as opposed to a Sakai search/browse/retrieve/edit tool for Fedora. I found a year-old document on Confluence with the title "Sakai Repository API Design" (http://bugs.sakaiproject.org/confluence/download/attachments/2852/repository-design-v1.doc?version=1). It offers two integration points for the content repository: a "Rep API" and a "JSR-170 API". I can find reference to JSR-170 as http://bugs.sakaiproject.org/jira/browse/REQ-210 but not reference to the Repository API. Help in navigating the Confluence and Collab sites would be most welcome. My thinking on the integration has evolved somewhat over the past few weeks (starting from the naive notion that providing an implementation for legacy/Entity.java class would be all that is required) to include these possibilities: * If the "Sakai Repository API Design" document of March 22, 2005, remains accurate, create a JSR-170 shim for Fedora (although, as the author of that document noted, "[Apache] Slide [a JSR-170 implementation] will not have explicit support for the Sakai organization structure...needed [for] efficient interrogation of access permissions..." Bridging access management structures will be a challenge, although I'm hoping (perhaps, again, naively so) that n-tier Shibboleth will help along these lines. * A Fedora shim that mimics the "Enterprise Repository Implementation" in the design document. * Or, most radically and improbable, an implementation of Hibernate over Fedora -- e.g., a object-relational-object storage structure. I'd like to try to firm up the possibilities within the next few days so as to include them on OhioLINK's Google Summer of Code idea list for developing proofs-of-concepts (if not actual robust implementations). (That list may have to include a suggestion for the JSR-170-to-Sakai shim, if that doesn't already exist...and judging from the Jira issue it doesn't.) Students will begin submitting applications on Monday, so it would be best if I can get it described by Friday. Thoughts? Peter -- Peter Murray http://www.pandc.org/peter/work/ Assistant Director, Multimedia Systems tel:+1-614-728-3600;ext=338 OhioLINK: the Ohio Library and Information Network Columbus, Ohio