This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
[S1S #030304_1 jdk1.4.1_02-b06] The 'Local Registry' menu item on RMI Registry node is disabled even if there is no running registry on the default port.
Created attachment 9292 [details] The NoClassDefFoundError.
The attached file is not for this bug! It's my mistake.
Raising priority to P1 as the bug is regression and without fix one RMI feature is not usable at all.
Fixed in [maintrunk]. I will also provide jar file for [nevada]/[release35] branches. Marek, could you please verify it?
Created attachment 9538 [details] rmi.jar for [nevada] and [release35] branches
Created attachment 9554 [details] Diff
Tomas, could you please review the fix? Thanks.
Short description of the fix: The bug is caused by changes in the openide module. However its probably easier to fix it within RMI module itself (although it does not solve the root cause of the problem). The issue is, that the LocalRegistryAction is having mode MODE_EXACTLY_ONE which means, that only exactly one node (and the one which is relevant for this action) must be selected to activate this action. However after changes in the openide module when selecting proper node (RMI Registry node in runtime) two nodes are always returned as active nodes: node itself and relevant filter node and therefore the action is always disabled. The easiest way to fix it is to change mode of the action from MODE_EXACTLY_ONE to MODE_ALL which will make the action enabled only when all activated nodes are relevant. And as there is only one real node which is relevant it should always produce proper result. Fix is very simple and should not have any negative impact on the RMI module or any other modules.
The patch seems fine. It fixes the problem does not affect the other module (IDE) behaviour.
Verified the bugfix in main trunk build #200303270100, with the attached module jarfile in netbeans 3.5 build #200303262350 and S1S Nevada build #030326.
approved for release35 by release coordinator
I'm not sure what changes in openide and what extra filter node you are referring to, but if you have reason to believe there is some regression in openide you had better file it (and make this issue depend on it). Looking at LocalRegistryAction, it is quite stupid - does not need to be a CookieAction at all. Should be a CallableSystemAction. That would be the better thing to do.
Jesse, I agree with you that the fix is not ideal and also I will file a bug against openide, however: 1. After my fix the bug does not depend on a openide fix as it will work anyway 2. We are in High Resistance mode so I think that its better to fix the problem using easy fix then most correct one. IMHO less chance of regression. Im closing it as fixed (its approved and also verified by QA). If you think that the action should be rewritten to be a CallableSystemAction, file new task.
OK, I filed #32467. Actually changing the superclass of the action should be quite trivial as well. BTW you forgot to set a target milestone on this issue.
Verified in Nevada 030331 build.