11 Dec, 2007

1 commit


10 Dec, 2007

1 commit

  • If the Replication Server is stopped while the ServerHandler writer threads are blocked waiting for some window credit they are blocked and don't stop
    immediately.
    
    This issue is minimal for real life but can cause unit test failure because replication Servers are stopped and started a number of times during the tests.
    
    This commit also enabled some more replication tests and change them to make the
    order in which they are run more predictible.
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3542 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     

08 Dec, 2007

2 commits


07 Dec, 2007

2 commits

  • The classes in org.opends.messages need to be made public so that they can be referenced from other public APIs. This change adds the PublicAPI annotation to the classes defined in this package as follows:
    
       * the org.opends.messages package is VOLATILE
    
       * the org.opends.messages core classes (Message, MessageDescriptor,
         MessageBuilder, Category, and Severity) are all UNCOMMITTED
    
       * the class org.opends.messages.MessageDescriptorRegistry is PRIVATE
         (it's only used in unit tests)
    
       * generated message classes are PRIVATE
    
    In addition minor cleanup was made to the classes where appropriate to make them suitable for public APIs:
    
       * used "final" where possible
    
       * reduced visibility of classes, methods, and members where appropriate
    
       * fixed some potential bugs spotted during this clean-up: class
         members not being initialized (the "final" keyword is useful for
         spotting this!)
    
       * added missing javadoc
       
       * I've removed the third-party and user-defined message generation
         since we could see know way in which they could be easily used
         without impacting core OpenDS source
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3536 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • dsreplication enable fails to initialise schema of the peer if the user schema
    contains a new objectClass dependent on a new attributeType.
    
    This was happening because the validation of the new objectClasses was done using the
    old schema while it must be done with the new schema currently being modified. 
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3532 41b1ffd8-f28e-4786-ab96-9950f0a78031
    gbellato
     

06 Dec, 2007

1 commit


05 Dec, 2007

3 commits


04 Dec, 2007

3 commits


03 Dec, 2007

1 commit

  • …d init/fini processing relevant to entry cache for all
      active entry cache implementations instead of letting each entry cache implementation to be BackendInitializationListener on its
      own which can lead to unexpected results such as listener activity on already finilized entry cache implementation.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3517 41b1ffd8-f28e-4786-ab96-9950f0a78031
    abobrov
     

02 Dec, 2007

1 commit


29 Nov, 2007

3 commits


27 Nov, 2007

2 commits


26 Nov, 2007

2 commits

  • The fix consists in including a little bit more on context to the error message.
    
    For command-line dsconfig list-local-db-indexes --backend-name tasks -n -p 1389 -w s
    
    The output was:
    
    "Task Backend" is not a recognized component type
    
    With the fix the output is:
    
    "Task Backend" component type cannot be used with subcommand
    {list-local-db-indexes} 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3497 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     
  • Fix for Issue #1961 (reviewed by Josu)
     * Remove instances of "Specifies the..."
     * Remove instances of "Indicates that..."
     * Reword instances of "Specifies to..." (which doesn't really make
       sense anyway)
     * Change instances of "This utility may be used..." to "This utility
       can be used..." as discussed in a previous issue
     * Change some erroneous instances of "may" to "might" or "can". This
       might sound like a pointless change but it has implications for
       localization - "may" is generally ambiguous. 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3496 41b1ffd8-f28e-4786-ab96-9950f0a78031
    lfrost
     

23 Nov, 2007

1 commit


22 Nov, 2007

4 commits

  • jcambon
     
  • The fix consists on creating a TopologyCache even if the ADS does not exist.  The code in TopologyCache will generate an exception describing the problem (if any). 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3487 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     
  • The problem is that the root server entry is not created by default (it is used by some parts of the code to know if the ADS was initialized or not).  The fix consists of updating the ADSContext.registerServer method to check if the parent entry exists.  The check is only applied here since this is a particular case (and by default this check is not required for other entries in the ADS). 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3486 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     
  • The upgrader is breaking replication (and not just dsreplication).
    
    The problem occurs because the upgrader is systematically overwriting the contents of the admin-backend.ldif file when it should not.  The fix consists on not updating this file unless it does not exist.
    
    The same fix has been applied to tool.properties since it was also systematically overwritten (and so the user lost the changes made to it).
    
    Appart some fixes related to message formatting are also included in the attached diff file.  The code used to display messages with HTML formatting in CLI mode, so specific messages have been added when the upgrade is run from the command line.
    
    Finally the code of the upgrader has been updated to use the MenuBuilder class.  This is done to have a consistent output with the rest of the command-lines. 
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3485 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     

21 Nov, 2007

2 commits


20 Nov, 2007

3 commits


19 Nov, 2007

3 commits


16 Nov, 2007

5 commits

  • This relates to issue 1829.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3462 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • There is a problem in windows when we launch scripts using Runtime.exec if the script path contains a '=' character.  The fix consists on quoting the script paths on windows when we launch them using Runtime.exec.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3461 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     
  • of component definitions as advanced and customizable:
    
    * advanced components: these will not be displayed by
      default in dsconfig. Examples include the various
      system backends
    
    * customizable components: these components are tagged
      as intended for customization using the CLI XML
      profile. Customizable components are usually the top
      level components previously referred to as "generic"
      by dsconfig. This term confused users who would, for
      example, mistake a "generic" backend for a JE backend.
    
    This change is part of issue 1829.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3460 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift
     
  • …the user already provided it.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3459 41b1ffd8-f28e-4786-ab96-9950f0a78031
    jvergara
     
  • various clean-up to the XML definitions:
    
       * clean up typos and any bad grammar in the synopsis/description
         elements
    
       * fixed common errors. For example, there were many places where the
         synopsis was repeated in the description (the description is only
         ever used in conjunction with the synopsis). There were also many
         places where the <adm:user-friendly-name> was being used in
         property definition description elements to refer to the property
         name, when it actually refers to the managed object name
    
       * split big synopsis (not sure of the plural form - synopsi?) where
         possible - there were lots of cases where we'd have a big synopsis
         but no description
    
       * overrode properties with sensible default values where
         appropriate (for example, I provided sensible default values for
         some of the attribute-type properties in the virtual attribute
         definitions)
    
       * generally made documentation more consistent across the definitions
    
       * removed unneeded XML attributes of the form 'mandatory="false"'.
         Attributes only need to be specified if their value is true as
         they default to false. Adding redundant attributes just makes the
         XML harder to read
    
       * removed any documentation that referred to the configuration in
         LDAP related terms (with one exception in the Schema backend). All
         documentation now refers to other properties using the property
         names, not their associated LDAP attribute.
    
    
    git-svn-id: https://svn.forgerock.org/opendj/trunk@3456 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew_swift