- 1ad7f69 Fixed Felix' delegation hack so that it correctly delegates to the parent by Richard S. Hall · 16 years ago
- 61ac34b Make overriding the built-in url handlers the default and only way. My last commit did remove the performance problems and it looks like this will become spec so we might as well already implement it (FELIX-756). by Karl Pauls · 16 years ago
- 6c906db Fix a small issue where relative urls would not be resolved when override is enabled. Furthermore, improve the performance of the stream handler proxy a bit. by Karl Pauls · 16 years ago
- 2cd1798 Applied patch (FELIX-799) to implement ServiceReference.compareTo() method. by Richard S. Hall · 16 years ago
- 1c0c784 Minor changes to comments. (FELIX-753) by Richard S. Hall · 16 years ago
- 58740f1 Improved some comments and other minor edits. (FELIX-753) by Richard S. Hall · 16 years ago
- 3bd1ca7 Switch URLHandlers to the old behavior of not overriding built-in handlers by default. Overriding can be enabled using a felix.service.urlhandlers.override=true system property. Furthermore, fix a bug that made URLHandlersBundleStreamHandler.openConnection(URL) fail when extending Felix. (FELIX-756, FELIX-800) by Karl Pauls · 16 years ago
- 43a1308 Make some members in the ldap operators final to hopefully fix toString() visibility issues. (FELIX-765) by Karl Pauls · 16 years ago
- 4ee838a Remove System.out debug message. by Karl Pauls · 16 years ago
- 15cf8c6 In certain cases it might be desirable to override built-in URLStreamHandlers with services provided by bundles. We will first see whether there is a service for the protocol and if so use that one else, we try to lookup a built-in one as we do now. This includes the possibility to add locations to the default locations. What changes is that we first look at the services. (FELIX-756) by Karl Pauls · 16 years ago
- 431f0b2 Fixed spec compliance issues around getting a service reference from an by Richard S. Hall · 16 years ago
- 001d6d2 Change the name of the SystemBundle interface to be Framework as a result by Richard S. Hall · 16 years ago
- 7225a21 Don't try to install extension bundles as fragments because they are different. Furthermore, fail to install extension bundles if they have incomplete metadata (FELIX-770). by Karl Pauls · 16 years ago
- 8ba9316 make both, the addToStreamCache and the addToContentCache methods synchronized. Then the sychronized block in both, the createUrlStreamHandler and createContentHandler methods can be narrowed down to protect the first line (FELIX-748). by Karl Pauls · 16 years ago
- 06487bd Some simple changes to framework. by Richard S. Hall · 16 years ago
- 69c0d3e Give bundles all permission if no security extension is installed (as per spec). by Karl Pauls · 16 years ago
- 8155b23 Modified framework to have default values for the system packages property, by Richard S. Hall · 16 years ago
- c0e2e61 Update felix to work with the latest version of the android sdk (1.0.0.-rc1). by Karl Pauls · 16 years ago
- 1032b7c Implemented remaining PackageAdmin methods from R4. (FELIX-35) by Richard S. Hall · 16 years ago
- ca964d7 Modified resolver to correctly mark fragment modules as resolved. (FELIX-777) by Richard S. Hall · 16 years ago
- 1e33c07 Minor code changes. by Richard S. Hall · 16 years ago
- d0fd653 Add missing import statement. (FELIX-776) by Richard S. Hall · 16 years ago
- 20310c4 Modified the resolver to longer consider uninstalled fragments and hosts by Richard S. Hall · 16 years ago
- 26bc9db Modified the resolver to sort candidates when resolving require-bundle by Richard S. Hall · 16 years ago
- a9aa21d Modified ExportedPackage.getImportingBundles() to include bundles by Richard S. Hall · 16 years ago
- 40f0fc5 Fix a NullPointerException in SecureAction when a SecurityManager is used by keeping pointers on the local stack. (FELIX-766) by Karl Pauls · 16 years ago
- 05a5de2 Null the classloader of a bundle when it's contenloader is closed and call System.gc twice when we can not delete the bundle revision in the cache. This allows to update and uninstall bundles with native libs on windows. (FELIX-733) by Karl Pauls · 16 years ago
- eeedd65 Moved removal pending flag from the module to the bundle. (FELIX-761) by Richard S. Hall · 16 years ago
- 8be364c Set the activator to null on Bundle.stop() to allow for earlier garbage collection. (FELIX-762) by Karl Pauls · 16 years ago
- 05ad584 More BundleInfo refactoring improvements. (FELIX-753) by Richard S. Hall · 16 years ago
- b6ed212 Implements support for flushing the cache on framework initialization. by Richard S. Hall · 16 years ago
- 16da5f4 The refactoring of BundleInfo was not complete, since RegularBundleInfo was by Richard S. Hall · 16 years ago
- 5a412b8 Applied patch (FELIX-750) to improve exception messages. by Richard S. Hall · 16 years ago
- 7e85cd8 The constructor to BundleInfo is not needed. (FELIX-753) by Richard S. Hall · 16 years ago
- 6ad0767 More changes for implementing framework API. (FELIX-753) by Richard S. Hall · 16 years ago
- f3ce002 Modified the bundle cache to no longer have profiles. Instead, it defaults by Richard S. Hall · 16 years ago
- d0c9b65 Modified the Felix API to aligned with the proposed standard OSGi framework by Richard S. Hall · 16 years ago
- ee13a05 Removed some unnecessary lines. by Richard S. Hall · 16 years ago
- b7c1424 Applied patch (FELIX-713) to support transiently starting/stopping bundles. by Richard S. Hall · 16 years ago
- 2fbab4b Modified the resolver so that it correctly discards exported packages from by Richard S. Hall · 16 years ago
- c0d7ad7 Refactored the resolver to clearly distinguish between "resolved" and by Richard S. Hall · 16 years ago
- 376e023 Do not bootdelegate in case Bundle.loadClass() has been used. (FELIX-712) by Karl Pauls · 16 years ago
- 00b9439 Remove extraneous import. by Richard S. Hall · 16 years ago
- 48934c5 Applied patch to decode reference URLs. (FELIX-731) by Richard S. Hall · 16 years ago
- d196b99 Applied patch (FELIX-728) to fix an issue where requests to Start Level by Richard S. Hall · 16 years ago
- 9137feb The class loader delegation in wires was not correctly working when a module by Richard S. Hall · 16 years ago
- 35a5498 Fixed a bug where class loader delegation for dynamic imports was happening by Richard S. Hall · 16 years ago
- 9ae129d Added a comment to remind us to clean up when fragments are done. by Richard S. Hall · 16 years ago
- e49bf1c Added a configuration property to determine whether installing a fragment by Richard S. Hall · 16 years ago
- 5085e43 Rollback an unintended commit from my previous commit. by Richard S. Hall · 16 years ago
- ad25073 Removed some experimental code for "implicit requirements" that was added by Richard S. Hall · 16 years ago
- e47c068 Fix an issue with FilterImpl.toString which could throw a NullPointerException when several Threads where doing a toString at the same time. The fix is making the children member of Operator volatile and assign it optimistic. (FELIX-721) by Karl Pauls · 16 years ago
- 4f7f1a7 Fixed a performance regression that was caused by uncached access to the by Richard S. Hall · 16 years ago
- 43f7228 Fixed a bug in calculating the system bundle's exported packages; it was by Richard S. Hall · 16 years ago
- a589b81 Support JAR-wide manifest package metadata. (FELIX-682) by Richard S. Hall · 16 years ago
- 367aae8 Throw an exception when installing a fragment that uses features that we by Richard S. Hall · 16 years ago
- 28397f6 There was a bug when calculating a bundle's class path for fragments, it by Richard S. Hall · 16 years ago
- d634ae8 Better error reporting when loading bundle classes using Bundle.loadClass. We now let the exception from ModuleImpl.loadClass bubble up instead of swallowing it. (FELIX-674). by Karl Pauls · 16 years ago
- 5ed610f More modifications for fragment support. Properly manages fragment by Richard S. Hall · 16 years ago
- b533765 We were not correctly checking attributes when resolving dynamic imports. by Richard S. Hall · 16 years ago
- 999d091 Implements singleton support, but we still need to test it more and perhaps by Richard S. Hall · 16 years ago
- 99bdda2 Subject.doAs is not considered by the spec and doesn't work well with the current approach we have inside the framework. This commit makes it work based on the current security policy installed. Furthermore, it is now possible (and needed) to assign permissions to bundles via the security policy (based on the bundle location - not certificates). In other words, as of now, correct permissions have to be assigned to the framework and bundles using the java security policy based on either the codesource or the subject, if subject.doAs is used. This is possible because bundles do have a codesource now and the bundle protection domain asks the installed policy for permissions in case no security provider is installed. (FELIX-654) by Karl Pauls · 16 years ago
- b00f75b Applied patch (FELIX-582) to return Bundle.getLastModified() for bundle by Richard S. Hall · 16 years ago
- 69b8073 Applied patch (FELIX-577) to improve handling of improper resource URLs. by Richard S. Hall · 16 years ago
- c6ae736 Add necessary stub methods to implement new R4.1 methods. (FELIX-595) by Richard S. Hall · 16 years ago
- 4402483 Applied patch (FELIX-650) to properly return symbolic name when it has by Richard S. Hall · 16 years ago
- 912b1f7 Fixed a bug where the m_revisions array was not being correctly rolled back by Richard S. Hall · 16 years ago
- 4a93fbf Modified manifest parser to be more lenient when reading the manifest version; by Richard S. Hall · 16 years ago
- 322c7d1 Adds support for the standard symbolic name for the system bundle; however, by Richard S. Hall · 16 years ago
- a6c0afc Applied patch to fix a type error when creating dynamic wires. (FELIX-606) by Richard S. Hall · 17 years ago
- 6aa170e Modified some comments. by Richard S. Hall · 17 years ago
- 448ab27 Initial patch to address fragment support. The included patch is only the by Richard S. Hall · 17 years ago
- 68429ac Minor code optimization: remove unnecessary null assignments. by Carsten Ziegeler · 17 years ago
- 569b156 Applied patch (FELIX-542) to give an error message when installing fragment by Richard S. Hall · 17 years ago
- 0eba918 Update NOTICE file to new layout, update documentation to match new style, delete redundant NOTICE and LICENSE files, prepare changelog for 1.0.4 release, and update dependencies to 1.1.0-SNAPSHOT. by Karl Pauls · 17 years ago
- 507d7e4 Check that the returned object from a service factory is an instance of all classes named when the service was registered (FELIX-540). by Karl Pauls · 17 years ago
- aa9b3a4 Reorder some instructions to speed up things a bit more. by Karl Pauls · 17 years ago
- 72dd265 The issue is that we log messages while holding framework internal locks -- hence, when a log service calls back into the framework (e.g., by loading a class) we might deadlock. For now we just disable logging to log services inside the framework. (FELIX-536) by Karl Pauls · 17 years ago
- 605c0b5 Cache filters and create an index for objectclass to improve service lookup performance. by Karl Pauls · 17 years ago
- 413d5b0 Don't use the info to get the bundle id but the bundle. Furthermore, add an index to get bundles by id. The reason is to speed up bundle lookup. by Karl Pauls · 17 years ago
- 270a9ae Reorder some statements to improve service lookup performance. by Karl Pauls · 17 years ago
- 1d26be3 Fix a spelling mistake. by Karl Pauls · 17 years ago
- f312b56 Cache the bundle id in the BundleImpl class to improve performance. by Karl Pauls · 17 years ago
- b5f4b62 Fix a spelling mistake by Karl Pauls · 17 years ago
- 25e5755 Make sure to catch NoClassDefFoundError when trying to query the framework classloader for whether a class is available or not. Previously, it was possible that this query was leaked to bundles because of the error propagate all the way to the bundle that instigated the classload. by Karl Pauls · 17 years ago
- ebce084 Don't try to resolve extension bundles because they are resolved at install time - this fixes an exception that could get thrown under certain conditions. by Karl Pauls · 17 years ago
- 5fbac03 Bundle.getResource does not check for AdminPermission, like getResources does. This adds the permission check. (FELIX-150) by Karl Pauls · 17 years ago
- 41db639 Make urlhandlers work when a handler factory is already set. Furthermore, it is now possible to have more then one framework running in more then one classloader. It is still possible to disable urlhandlers per framework. If no framework has urlhandlers enabled the factory will not be set. (FELIX-38). by Karl Pauls · 17 years ago
- 3a7f747 Add some more methods to SecureAction that allow for reflection and unsetting URLStreamHandlerFractory. by Karl Pauls · 17 years ago
- 08b2ee6 Make extension manager use internal api to minimize overhead. by Karl Pauls · 17 years ago
- 020dc14 Make the framework check for valid bundles on install as per spec. by Karl Pauls · 17 years ago
- 758c2be Make the framework check for valid bundles on install as per spec. by Karl Pauls · 17 years ago
- 2e34f03 Use a new method on pd to check for allowed exports. by Karl Pauls · 17 years ago
- d1e209b The extensionmanager creates a classcastexception if more then framework is around and both have extensions installed. This fixes the issue. by Karl Pauls · 17 years ago
- ed61fa5 Add a method to the external interface and fix hashcode, equals, and hashcode. by Karl Pauls · 17 years ago
- 9794cc2 Turns out our LDAP filter is not reentrant - but it is easy to fix - so this patch should make it work. It is needed because filter evaluations can call out to keys being compared which might in turn use the given thread to evaluate their own filter. by Karl Pauls · 17 years ago
- 9b2a329 Null a local variable to help speed garbage collection. by Richard S. Hall · 17 years ago
- 5eda19c Refactored the IContent abstraction from the module loader layer to provide by Richard S. Hall · 17 years ago
- 8c0e750 Applied patch (FELIX-500) to disable the class loading diagnostic message by Richard S. Hall · 17 years ago
- 71ca406 Modified the service registry to use more fine-grained locking to avoid by Richard S. Hall · 17 years ago