Return-Path: Delivered-To: apmail-incubator-felix-dev-archive@www.apache.org Received: (qmail 90300 invoked from network); 24 Feb 2006 23:09:27 -0000 Received: from hermes.apache.org (HELO mail.apache.org) (209.237.227.199) by minotaur.apache.org with SMTP; 24 Feb 2006 23:09:27 -0000 Received: (qmail 47808 invoked by uid 500); 24 Feb 2006 23:09:26 -0000 Delivered-To: apmail-incubator-felix-dev-archive@incubator.apache.org Received: (qmail 47775 invoked by uid 500); 24 Feb 2006 23:09:26 -0000 Mailing-List: contact felix-dev-help@incubator.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: felix-dev@incubator.apache.org Delivered-To: mailing list felix-dev@incubator.apache.org Received: (qmail 47764 invoked by uid 99); 24 Feb 2006 23:09:26 -0000 Received: from asf.osuosl.org (HELO asf.osuosl.org) (140.211.166.49) by apache.org (qpsmtpd/0.29) with ESMTP; Fri, 24 Feb 2006 15:09:26 -0800 X-ASF-Spam-Status: No, hits=1.4 required=10.0 tests=DNS_FROM_RFC_ABUSE,HTML_10_20,HTML_MESSAGE,SPF_PASS X-Spam-Check-By: apache.org Received-SPF: pass (asf.osuosl.org: domain of Jeff_McAffer@ca.ibm.com designates 32.97.182.141 as permitted sender) Received: from [32.97.182.141] (HELO e1.ny.us.ibm.com) (32.97.182.141) by apache.org (qpsmtpd/0.29) with ESMTP; Fri, 24 Feb 2006 15:09:25 -0800 Received: from d01relay02.pok.ibm.com (d01relay02.pok.ibm.com [9.56.227.234]) by e1.ny.us.ibm.com (8.12.11/8.12.11) with ESMTP id k1ON93hB019918 for ; Fri, 24 Feb 2006 18:09:03 -0500 Received: from d01av02.pok.ibm.com (d01av02.pok.ibm.com [9.56.224.216]) by d01relay02.pok.ibm.com (8.12.10/NCO/VERS6.8) with ESMTP id k1ON94Af205582 for ; Fri, 24 Feb 2006 18:09:04 -0500 Received: from d01av02.pok.ibm.com (loopback [127.0.0.1]) by d01av02.pok.ibm.com (8.12.11/8.13.3) with ESMTP id k1ON94pQ030461 for ; Fri, 24 Feb 2006 18:09:04 -0500 Received: from d25ml04.torolab.ibm.com (d25ml04.torolab.ibm.com [9.26.6.105]) by d01av02.pok.ibm.com (8.12.11/8.12.11) with ESMTP id k1ON94oe030452 for ; Fri, 24 Feb 2006 18:09:04 -0500 In-Reply-To: <200602242306.59396.marrs@xs4all.nl> To: felix-dev@incubator.apache.org MIME-Version: 1.0 Subject: Re: Issues with installing bundles by reference X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 From: Jeff McAffer Message-ID: Date: Fri, 24 Feb 2006 18:08:48 -0500 X-MIMETrack: Serialize by Router on D25ML04/25/M/IBM(Release 6.5.4|March 27, 2005) at 02/24/2006 18:09:05, Serialize complete at 02/24/2006 18:09:05 Content-Type: multipart/alternative; boundary="=_alternative 007EB25A8525711F_=" X-Virus-Checked: Checked by ClamAV on apache.org X-Spam-Rating: minotaur.apache.org 1.6.2 0/1000/N --=_alternative 007EB25A8525711F_= Content-Type: text/plain; charset="US-ASCII" Marcel Offermans wrote on 02/24/2006 05:06:59 PM: > The first scenario I thought of was when developing and deploying bundles, > being able to actually have bundles that reference a directory structure > would eliminate the need to (re)build bundle jars (and after triggering an > update in the framework, having that framework extract them again). This > scenario seems to be used in J2EE appservers too (for .war and .ear files). Certainly this is an important one however it actually requires more than just install by reference. For example, if you have some project (whatever) of source and you build it and want to install the result by reference as a bundle, the result has to be in the shape the bundle describes. In particular, the Bundle-Classpath has to correctly identify the location of hte classes and resources. Since the common case is to have a classpath of . then it means that the package dirs for the classes have to be directly in the parent of the MANIFEST.MF file or you have to implement a mechanism that allows you to map the classpath elements to build locations. The latter is what we do in Eclipse to enable PDE's direct launching of plugins/projects. This allows people to structure their output locations however they want and still just write code, save and launch. No explicit compile or build steps needed. > What other scenario's did you have in mind (if any)? The other main usecase we have for by-reference is scaleablity. If your system is actually installed on disk (as opposed to downloaded dynamically) then installing bundles by value duplicates the storage required. Further, if you have the same disk-based bundle in multiple configurations/profiles, you would have to have one copy for each. The approach we have taken is to allow for pools of bundles to be placed on disk and then define configurations that identify the plugins from that pool that should be run together. No duplication, no copying, ... Since some of the systems built on Eclipse are >1GB, this is an obvious win. Jeff --=_alternative 007EB25A8525711F_=--