Datasets:

AndressaStefany

/

bug-reports

like 0

{'$oid': '52e9b48d54dc1c25ebdb20d9'}

1. In a suspended thread, go to the editor, type in 1+1 anywhere. 2. Revert Observe: The method you were in is lost. You have to reselect the corresponding stack frame. NOTES: DS (8/3/2001 3:22:13 PM) Really an instruction pointer issue.

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b48d54dc1c25ebdb20da'}

DW (6/6/01 4:26:25 PM) On Win98, stepping in an attach launch is slow. Investigate. JGS & DS (6/7/01 5:20:09 PM) We tried multiple combinations of server and client: - Server pgm running on 98, client attach launch on 98, stepping very slow - Server pgm running on NT, client attach launch on 98, stepping normal - Server pgm running on 98, client attach launch on NT, stepping very slow - Within this scenario, we tried server pgm running under JDK122 & IBM VM, both were very slow Tentative conclusion: Something about 98 (sockets? TCP/IP stack?) makes attaching to a running program on 98 very slow. Suggest this be made into a README item, e.g., "Attaching to a program running on 98 can result in sluggish response in the debugger". DW (6/15/01 10:54:09 AM) Added to README.

normal

CLOSED

WONTFIX

{'$oid': '52e9b48d54dc1c25ebdb20db'}

When running a program under debug from the Type Hierarchy perspective, the console is opened in this perspective. This is inconsistent with running the program from the Java perspective where the console of the Debug perspective is used. NOTES: DW (6/6/01 12:00:46 PM) If you have the debug preference set to "Show console when there is program output" and you are in a non-debug perspective, a console will open. JBL (6/6/2001 7:06:25 PM) But why isn't it shown in the Debug perspective and the Debug perspective given focus? DW (6/6/01 12:21:12 PM) We only auto-switch to debug perspective on a launch or when a thread suspends. Otherwise, we let the user change to any perspective they want (and thus do not auto-switch back when there is console output). This could be another (future) preference.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b48d54dc1c25ebdb20dc'}

DS (6/5/01 1:08:01 PM) Test Case Self hosting Hit a breakpoint The toolbar actions and debug menu items are not enabled for the selected stack frame. Reselecting the stack correctly enables the actions and items. DW (6/11/01 8:09:18 AM) Not critical - reselecting fixes the problem. Does not happen consistently. DW (8/14/01 4:00:12 PM) Is this still a problem? (with debug UI re-work)

minor

VERIFIED

FIXED

{'$oid': '52e9b48d54dc1c25ebdb20dd'}

Since the debugger shows only one method at a time and this method is the one of the selected stack frame, I find the Outline on the right useless. If I want to browse code, I would go to the Java perspective. NOTES:

normal

RESOLVED

WONTFIX

{'$oid': '52e9b48d54dc1c25ebdb20de'}

DW (6/7/01 9:27:33 AM) 119++ When doing an eval in the scrapbook/debugger, the package "org.eclipse.jdt.internal.eval.target" appears in the workspace, which is the home of "CodeSnippet". We do not clean-up (remove) this package when the target termiantes. DW (6/7/01 2:09:06 PM) Simple fix would be to have the snippet support change the "CodeSnippet" class to be in the default package. There is a nasty side effect here. When a project uses the defualt source and binary folder config (i.e. project is the root), a new package appears when a evaluation is done. This package will also show up when releasing to a repository. This problem does not surface in the self hosting world because the package is hidden in the output directory. Another potential fix is to have the debugger delete the package - but that should only be done if the package is empty. Propose fix in JCORE. DW (6/8/01 8:54:09 AM) Make into a README. DW (6/15/01 10:08:55 AM) Added to README.

critical

VERIFIED

FIXED

{'$oid': '52e9b48d54dc1c25ebdb20df'}

I changed the contents of the method with the breakpoint BUT did not save them when I closed the debugger. I debugged the same test which hit the same breakpoint... the unsaved changes were still there. See 1GELY07: ITPDBG:ALL - Should last editor remain open after process is terminated NOTES: DW (6/7/01 11:36:35 AM) This is under editor manager control. Not a debugger issue.

normal

CLOSED

INVALID

{'$oid': '52e9b48d54dc1c25ebdb20e0'}

DW (6/7/01 3:15:10 PM) Thread termination is not supported on JDK 1.3. The debugger implements thread termination by invoking #stop on a thread with an instance of ThreadDeathException. The debugger attempts to create an instance of ThreadDeathException on the first class load event received from the VM. On 1.3, the ThreadDeathException consrtuctro throws a NullPointerException when invoked before a program reaches "main" (not sure why). Users should beware not to catch NullPointerException or any of its supertypes until after a program has reached main, as this causes the attempt to create the ThreadDeathException to suspend, from which the debugger may not recover. DW (6/15/01 10:42:28 AM) Added to README.

enhancement

RESOLVED

WONTFIX

{'$oid': '52e9b48d54dc1c25ebdb20e1'}

DS (6/7/2001 11:00:13 AM) 119++ on Linux In the processes view, if you terminate a process from a debug launch, the launch does not update to show as terminated. It does show as terminated in the debug view. DS (6/8/01 11:10:13 AM) 120++ not a problem on NT. DS (6/8/2001 12:38:43 PM) A problem on linux 120 running the out of the box jre 1.3 DW (6/12/01 8:47:19 AM) Deferred.

normal

VERIFIED

FIXED

{'$oid': '52e9b48d54dc1c25ebdb20e2'}

DW (6/7/01 3:31:56 PM) Terminating an IBM 1.3 VM on Linux does not work properly. Terminating the launch or the associated OS process has the desired effect. When a debug target (VM) is selected in the debug view and terminate is pressed, the VM will remain in with all threads. The workaround is to use "terminate and remove". There appears to be a bug in the JDI (target side) implementation or VM. This only occurrs when a thread in the VM being terminated has a suspended thread. It appears the VM waits forever for the suspended thread to complete. DW (6/8/01 12:49:34 PM) It appears that terminating the suspended thread will clean things up since that also resumes the thread (so it can terminate), which allows the VM to exit. DW (6/15/01 10:58:28 AM) Added to README.

normal

RESOLVED

INVALID

{'$oid': '52e9b48e54dc1c25ebdb20e3'}

DS (6/8/01 10:20:19 AM) When you attempt to do a Display from a JavaEditor within an inner type or binary type, the selected text swiped for evaluation will still appear in the display view. The display will fail with an error dialog stating "Unable to perform evaluation in an inner type" When you attempt to do a Display from the display view, in the context of an inner type or binary type stack frame, a semi colon will be added to the display view. The display will fail with an error dialog stating "Unable to perform evaluation in an inner type" DW (6/15/01 10:59:16 AM) Added to README.

normal

CLOSED

WONTFIX

{'$oid': '52e9b48e54dc1c25ebdb20e4'}

On the first launch of a class, I always get the Select Launcher dialog which offers the choice of selecting only 1 launcher. Maybe we should skip this part if there is only 1 choice. Same would apply to the Select Target part. When I select a class and press run, there is only 1 possible choice: the class I selected. NOTES: DW (6/12/01 8:35:38 AM) Deferred. DW (8/8/01 2:49:43 PM) Duplicate of 1GCSYAF: ITPDUI:ALL - Setting default launcher with launch element selected DW (8/10/01 4:22:07 PM) Issues in JavaApplicationWizardPage * it uses an asyn exec to update the page on entry -> should use setVisible(true) * message update should be based on selection, not on page entry

minor

RESOLVED

DUPLICATE

{'$oid': '52e9b48e54dc1c25ebdb20e5'}

- new Java Project - new Scrapbook page - use toolbar package button to have it run in java.util - evaluate: new Hashtable() - you get: An exception occurred during evaluation: java.lang.SecurityException NOTES: NE (6/12/01 11:14:23 AM) I noticed that it put a java.util package in my project. I guess the VM doesn't let you shadow JDK packages. EG (6/12/2001 5:18:04 AM) this has worked in the old evaluation support. EG (6/12/2001 5:14:57 AM) this has worked in the past. Run in works for application packages, e.g., I can do run in "junit.text.ui" and evaluate new TestRunner(). Suggest README Moving to DUI JGS (6/12/01 10:38:06 AM) Can duplicate under 122. This problem was originally filed as 1GEWPXV: ITPDUI:ALL - SecurityException running in package but was closed because it couldn't be duplicated under 119. It appears to be back with 122. DW (6/12/01 12:29:00 PM) Problem does not occur with Sun JDK 1.2.2 or J9. README. DW (6/15/01 10:12:57 AM) Added to README.

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20e6'}

Old system: We deal only with IMarkers. We can't define our own IMarkers (either by implementing the interface or subclassing the implementation), so we just use them directly. We're implementing the "concept" of breakpoints into the system without a breakpoint class. This is done by adding attributes to IMarkers and doing processing based on the value of these attributes. New system: We add the IBreakpoint interface to the system and go about implementing breakpoints in one of two ways: 1) We subclass the IMarker implementation (Marker) with the contract that all things related to persistence will remain the sole domain of the Marker superclass. This way, the Breakpoint class "is a" Marker. 2) We aggregate an IMarker. This leaves the existing contract in place, but adds an extra level of indirection in that the Breakpoint implementation will constantly be passing messages to its IMarker instance instead of just calling them on itself. This way, the Breakpoint class "has a" Marker. Thanks to the nature of objects, either of these implementation choices will work without affecting the objects that deal with IBreakpoints. The following analysis holds regardless of what implementation we choose. Good things about the current system: 1. It got the job done in a development cycle that was hampered by uncontrollable external forces. Problems with the current system : 1. Inappropriate Intimacy - Processing and data storage related to breakpoints are spread out across the system. 2. Procedural Code Masquerading as Objects - The DebugJavaUtils "object" is the biggest offender here. It is the closest thing to centralization of breakpoint- marker processing, but it is nothing but a large clump of procedures. It is a frequently used "object" in the system but it satisfies none of the goals of OO and breaks most of the rules. 3. Feature Envy - Again, DebugJavaUtils is the biggest offender here. All of its procedures operate on someone else's data (IMarker). 4. Duplicate Code - Code for accomplishing the same thing for different types of breakpoints is frequently repeated. In some cases, code that is identical, or nearly so, is defined for each marker type. 5. Switch Statements - The code is filled with method dispatch based on the IMarker type. Methods must frequently dispatch to specialized (and uniquely named) methods for doing the same thing for each different type of marker. Example 1. A method called addBreakpoint(IMarker marker) defined in the JDIDebugTarget manually checks the type of the marker and then dispatches to addMethodEntryBreakpoint(marker), addExceptionBreakpoint(marker), addLineBreakpoint(marker), etc. 6. Adding a new kind of breakpoint in the current system requires changes (adding dispatches and breakpoint_type specific methods/code) all over the system. Only someone who knows the system inside and out can know all of the places to make these changes. Good things about the new system: 1. Encapsulation - All data and processing related to breakpoints are centralized in the IBreakpoint hierarcy. Removes problems 1, 2, and 3. 2. Inheritance - The new system allows breakpoints to make use of state and behavior defined in superclasses. This means that behavior which is shared by a superclass and its subclasses is defined Once And Only Once. Removes problem 4. 3. Polymorphism - The manual dispatching which litters the code is eliminated through polymorphism. Example 1 is refactored with double dispatch as follows: Example 2 The method addBreakpoint(IBreakpoint breakpoint) calls breakpoint.addToTarget(this). Removes problem 5. 4. Extensibility - Adding a new breakpoint class is easy to do. The implemented interface (IBreakpoint) and/or inherited classes clearly specify exactly what methods need to be written and once they're written the new breakpoint type will just go. Thanks to polymorphism the rest of the system doesn't need to know about new breakpoint types. Removes problem 6. Bad things about the new system: 1. Requires some API parameter changes from IMarkers to IBreakpoints. Some specific examples are IBreakpointManager and IDebugTarget. NOTES: DW (8/13/01 2:30:15 PM) Breakpoint definition: To define a breakpoint, clients will need to define a marker extension, that documents the attributes that their breakpoint marker will have. In addition, a "breakpoint" extension will correspond to the marker definition, defining the "marker type" that the breakpoint corresponds to, as well as the fully qualified name of the class that implements the breakpoint (an implementation of IBreakpoint). Breakpoint creation: Breakpoints are created atomically with a corresponding marker, and added to the breakpoint manager. Breakpoint change notification: The IBreakpointSupport interface will be eliminated. IBreakpointListener will remain, but IDebugTarget will no longer be required to implement it. Any clients interested in breakpoints must listen for changes. Editors will be able to listen to marker changes to update. (Any change in a breakpoint must be reflected in its marker in order for the UI to update properly). The JDIDebugTarget (implementation) will be a breakpoint listener, and will delegate to breakpoints as added/changed/removed.

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20e7'}

DS (6/11/2001 11:24:37 AM) In the debug view, select a debug target Using the debug pulldown menu to select the Remote Java Application Launcher The Attach wizard comes up, but filling it in with valid data, and hitting Finish results in nothing happening and the wizard not closing. DW (6/12/01 12:40:38 PM) Deferred. Workaround is to select a non-debug element. DW (6/15/01 11:03:18 AM) Added to README.

critical

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20e8'}

In the variables view, move the mouse over the Show hexadecimal values, Hide static fields Hide final fields Note that the rollover image is the same as before rollover (no color) NOTES: DW (6/11/01 1:52:23 PM) These actions only specify their images in plugin xml, as they are contributed to the view via an extension point. Extension point does not allow for enabled, disabled, and hover images. DW (6/12/01 8:36:18 AM) Deferred. JGS (8/10/01 2:53:01 PM) Waiting on 1GHUL80: ITPUI:WIN2000 - Contributed actions should have hover and disabled images

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20e9'}

DS (6/11/2001 11:59:55 AM) Test Case Debug ConsoleTest Enter the required input Relaunch last Go to the console of the first launch Swipe all, select copy from the context menu Go to the console of the newest launch The paste action is not enabled. DW (6/13/01 10:29:10 AM) Defer.

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20ea'}

After playing around with 1GF7QYW: ITPJUI:WIN2000 - SecurityException when running scrapbook in java.util I was left with a java.util package in my project which couldn't be deleted. The delete looked like it did something, but it was still left in the workspace. I was trying to delete it from the packages view. It deleted OK when I tried it from the Navigator view. Couldn't reproduce. NOTES: EG (6/12/2001 5:50:53 AM) moving to DUI DW (6/12/01 11:20:44 AM) Known problem. The debugger leaves directories in the output location of the project (which could overlap with source area). Deferred.

enhancement

RESOLVED

DUPLICATE

{'$oid': '52e9b48e54dc1c25ebdb20eb'}

When I inspect or display an expression in the debugger, it always resets the selection to be the line for the current instruction pointer. It should leave the selection alone. NOTES: DS (7/12/01 2:12:10 PM) This will be better addressed if we move to using an instruction pointer instead of relying on selection to show the current line of execution.

enhancement

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20ec'}

AK (6/12/01 5:12:37 PM) 1. junit 2. breakpoint in the first line of VectorTest::setUp 3. debug 4. hit the breakpoint 5. select fFull in the method body (in the editor and use inspect) 6. the correct value i.e. null shows up 7. step over, 8. step over note: the value of fEmpty is not updated ! 9. select it again and use inspect again 10. now it does show the updated value NOTES: DW (6/12/01 10:22:20 AM) This is the expected behaviour. When you inspect "fFull", you are evaluating the expression "fFull", which evaluates to null. You are NOT inspecting the field. After fFull has been assigned a value, the updating works because the value associated with the expression "fFull" is an object that is changing. Make a README for June. DW (6/15/01 11:13:55 AM) Added to README.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b48e54dc1c25ebdb20ed'}

How do I debug code in the scrapbook ? Do we have a halt and inspect methods a I can use? NOTES: EG (6/12/2001 8:54:02 AM) no halt but you can catch exceptions and throw them from a snippet ->README

minor

RESOLVED

INVALID

{'$oid': '52e9b48e54dc1c25ebdb20ee'}

Create a scrapbook page. Type "System.out.println("fooo");" Select it and run a dozen of times. 12 CodeSnippet_xxx.class will be created. If you keep using that page, it will create more and more CodeSnippet_xxx.class. When are these files deleted? It also creates a forder org.eclipse.jdt.internal.eval.target and adds a .class file in that folder. Couldn't it be in the same folder as the .jpage if the user does not specify a run in package. NOTES: EJP (6/12/01 2:35:06 PM) If you delete all those temp files your self you get a NoClassDefFoundError next time you try to use that page and the page becames useless. EJP (6/12/01 2:38:32 PM) If you terminate the page the CodeSnippet_xxx.class are deleted but the org dir is not. EG (6/12/2001 8:55:14 AM) moving to DUI - suggest README DW (6/15/01 10:14:38 AM) Added to README.

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20ef'}

DW (10/10/2001 9:59:28 AM) * Flooding the conole with output that does not have "new lines" causes the UI to become unresponsive. * Running the same program with "println" instead of "print" seems to work OK. public class Flood { public static void main(String[] args) { while (true) { System.out.print("junk"); } } } DW (10/10/2001 1:38:24 PM) Can we tell if the console is being flooded, and slow down the rate at which we write output to the console, giving more cycles to the UI? DS (10/10/2001 1:57:20 PM) Duplicate of 1GFBNFP: ITPDBG:WIN2000 - Eclipse Crash: console view

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b48e54dc1c25ebdb20f0'}

When the display fails to resolve an expression, the client prints "<inspection target>: " as usual, but it does not print a result to the display. Suggest printing "(failed to resolve)" the way "(no explicit return value)" is printed for declarations. 1. In the debugger display window, enter "int i=0;", highlight, and display. Client prints "(no explicit return value)" 2. Type a nonexistant variable name, highlight, and display. Client currently prints nothing. NOTES: DW (6/13/01 11:36:46 AM) Defer.

enhancement

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20f1'}

DS (6/12/2001 4:10:02 PM) Running self hosting workspace on IBM1.3 vm Log: Wed Jun 13 15:05:10 GMT+00:00 2001 2 org.eclipse.core.runtime 2 Problems occurred when invoking code from plug-in: org.eclipse.core.runtime.java.lang.NullPointerException at org.eclipse.jdt.debug.core.JDIDebugModel.getPluginIdentifier(JDIDebugModel.java:323) at org.eclipse.jdt.internal.debug.core.JDIDebugElement.targetRequestFailed(JDIDebugElement.java:291) at org.eclipse.jdt.internal.debug.core.JDIDebugTarget.terminate(JDIDebugTarget.java:1449) at org.eclipse.debug.core.Launch.terminate(Launch.java:225) at org.eclipse.debug.internal.core.LaunchManager.shutdown(LaunchManager.java:222) at org.eclipse.debug.core.DebugPlugin.shutdown(DebugPlugin.java:189) at org.eclipse.core.internal.plugins.PluginRegistry$2.run(PluginRegistry.java:253) at org.eclipse.core.internal.runtime.InternalPlatform.run(InternalPlatform.java(Compiled Code)) at org.eclipse.core.internal.plugins.PluginRegistry$1.visit(PluginRegistry.java:262) at org.eclipse.core.internal.plugins.PluginRegistry.accept(PluginRegistry.java:41) at org.eclipse.core.internal.plugins.PluginRegistry.shutdownPlugins(PluginRegistry.java:265) at org.eclipse.core.internal.plugins.PluginRegistry.shutdown(PluginRegistry.java:240) at org.eclipse.core.internal.runtime.InternalPlatform.loaderShutdown(InternalPlatform.java:467) at java.lang.reflect.Method.invoke(Native Method) at org.eclipse.core.internal.boot.InternalBootLoader.shutdown(InternalBootLoader.java:883) at org.eclipse.core.internal.boot.InternalBootLoader.run(InternalBootLoader.java:815) at org.eclipse.core.boot.BootLoader.run(BootLoader.java:280) at java.lang.reflect.Method.invoke(Native Method) at org.eclipse.core.launcher.Main.basicRun(Main.java:126) at org.eclipse.core.launcher.Main.run(Main.java:408) at org.eclipse.core.launcher.Main.main(Main.java:281) DW (6/12/01 4:18:23 PM) This problem occurs because of a known problem with terminating a suspended VM. See: 1GEYWL7: ITPDUI:ALL - README: Terminating VM on Linux There is no good workaround to this problem. Suggest as README. DW (6/13/01 10:31:46 AM) Fix is to change the code that retrieves the plugin identifier to use a constant instead of asking the plugin for an identifier. Alternatively, the JDI debug plugin could terminate its launches when shutdown. There is an interesting problem here - the JDI debug model has already been shut down, and now the generic debug plugin is being shutdown, which terminates all launches. There are instances of a debug model that has already been shutdown. Should each debug model be responsible for terminating its launches when shutdown? DW (6/13/01 3:05:55 PM) Suggest to defer. Side effect is that not all launches will be terminated when the IDE shuts down. DS (7/10/01 11:58:11 AM) Created constant plugin identifier for the JDI debug model. ARCH (7/11/01 12:09:00 PM) Fix it but not sure the fix proposed is right. Additional investigation needed DW (7/17/01 7:34:58 PM) We are no longer able to duplicate this problem. When running in debug mode, and the IDE is exited (on Linux), with a debug target active which has a suspended thread, the IDE will exit, and the debugger will dump an error to system out saying that it was unable to terminate a debug target - which is the expected behaviour. We no longer get the unexpected walkback. On investigation we found that the debug plugin attempts to terminate the system process associated with the debug target first (which fails). It then tries to terminate the debug target, which accepts the request without error. However, the debug target never gets a "VM Disconnected" callback, and thus it does not set its state to terminated. Previously, to get the error in this PR, the debug target was failing when we asked it to terminate. Propose no fix/change.

critical

RESOLVED

WONTFIX

{'$oid': '52e9b48e54dc1c25ebdb20f2'}

I set a breakpoint and launched eclipse in debug. Hit the breakpoint. Stepped over the code, later resumed. Hit the breakpoint again. Removed the breakpoint and hit resume. The breakpoint was triggered every subsequent time; i.e. the breakpoint marker (UI element) was not shown, but the breakpoint was still triggered. NOTES: DW (6/13/01 2:00:33 PM) Has been reported before, but never with a reproduceable case. Suggest defer.

normal

RESOLVED

INVALID

{'$oid': '52e9b48e54dc1c25ebdb20f3'}

Place a break point in some class and run it so that you have a thread suspended in the debugger. Open up some other .java files Select text in one of the .java files Note that in the editor of this other file, Inspect, Disply and Run to line are enabled. Likewise in all the .java editors Note that Inspect and Display generally do not do much, typically just stating that the field name could not be resolved. But run to line triggers an action, it causes the selected/suspended thread to run even if you are not in the right editor. NOTES: EG (6/14/2001 7:34:42 AM) the enabling state currently only checks that there is a debugger context. Not critical moving to DUI. DW (6/14/01 12:50:53 PM) Deferred. DW (8/15/01 9:55:57 AM) Duplicate of 1GF7DEK: ITPJUI:WINNT - contextmenu entries do not disappears in editor after snippet evaluation (Closed Dup).

minor

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20f4'}

EG (6/15/2001 12:36:46 AM) method entry breakpoints show up at the first line of the Java doc comment. The method entry breakpoint should better show up at the method declaration itself. NOTES: DW (6/15/01 9:05:02 AM) Defer.

normal

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20f5'}

jkca (6/14/2001 5:32:13 PM) jre-sdk 123 I had one run launch with console output and two debug launches with no output. The first debug launch terminated, so I clicked remove all. This removed the terminated debug launch, and set the console output to be that of the run launch. This was confusing to me because I expected the console to get set to the output of the remaining debug launch (and I had forgotten that I had a run launch sitting around). DW (6/14/01 5:51:35 PM) Defer.

normal

RESOLVED

WORKSFORME

{'$oid': '52e9b48e54dc1c25ebdb20f6'}

When there is an infinite flood of output without LF/CR to the console view, Eclipse crashes. The problem occurs when the horizontal scroll bar grows infinitely. Note that there is no problem if there are occasionally LF/CR in the output. To reproduce, run the following sophisticated code public class Main { public static void main(String[] args) { while( true ) System.out.print( "Britney Spears rulez" ); } } NOTES: DW (6/14/01 2:58:36 PM) Known problem. The IDE did not crash when I tried this. However, I had to kill the test program from the OS, and the IDE took a long time to catch up. Deferred. DS (9/28/2001 3:32:22 PM) Can we try to keep track how fast stuff is coming in and then "slow" it down if it is a flood. DS (9/28/2001 5:29:23 PM) Seems to work pretty well. Code like this in ConsoleDocument#streamAppended long currentTime= System.currentTimeMillis(); if ((currentTime - fLastStreamAppendedTime) < 400) { try { //slow down the flood of stream text to the console Thread.currentThread().sleep(150); } catch (InterruptedException ie) { } } fLastStreamAppendedTime= currentTime;

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b48e54dc1c25ebdb20f7'}

DS (6/27/2001 2:48:29 PM) If it is possible to determine if the current thread is paused, we should not be writing to standard in.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b48e54dc1c25ebdb20f8'}

I had to debug a client-server app and found some problems: - when running, I don't see which program/process currently writes to the console (if I don't use the Debug view) - even when running in the Debug view, I have to manually switch the console to see what happens (e.g. looking at simple sample as ping-pong) Workaround: I used (detached) the consoles from two different Debug perspectives. NOTES: DW (7/16/01 8:33:37 AM) Can we change the title of the console view to show the process name?

enhancement

VERIFIED

FIXED

{'$oid': '52e9b48e54dc1c25ebdb20f9'}

DW (8/14/01 9:35:25 AM) @see 1GD28BZ: ITPDUI:WINNT - Second attach to J9 VM shows "Debug response server" as paused Once the above PR is fixed, we will need to initialize thread status based on the JDI thread status.

minor

CLOSED

WONTFIX

{'$oid': '52e9b48f54dc1c25ebdb20fa'}

DS (7/2/01 10:18:22 AM) Test Case Point a breakpoint in Hello where indicated, and a breakpoint in the toString() Debug Swipe new Hello() line and Display You correctly timeout evaluating the toString, and the debug view but for some time the entire debug view is blank and then redraws. Code: public class Hello { public static void main (String[] args) { String out= "Hello"; new Hello(); System.out.println("Breakpoint here"); } public String toString() { return "Another Breakpoint"; } } DS (8/9/01 1:37:36 PM) NE and I looked into the "flash" problem in the viewers. I believe that he is looking into the problem. May just be an underlying widget limitation.

minor

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb20fb'}

Created a project with classpath containing project org.eclipse.jdt.core in its classpath. Trying to activate the eclipse batch compiler in a snippet does not work and raises a dialog saying "Evaluation failed: internal error - unable to obtain an execution context". Nothing in the .log. org.eclipse.jdt.internal.compiler.batch.Main.compile( "N:/WWW/numbat/Testing/JCK-13/JCK-compiler-13a/tests/lang/ICLS/icls042/icls04214/icls04214.java " +"-classpath d:/jdk1.3/jre/lib/rt.jar") NOTES: PM (7/10/2001 12:06:23 PM) Actually, evaluating '1+1' did not work either, same behavior. EG (7/11/01 11:55:27 AM) I can evaluate 1+1 and it worked. Which JRE are running on? Is the 1+1 failure a follow-up of the failed evaluation of the batch compiler? PM (7/11/2001 11:54:18 AM) Running with Sun 1.3. It isn't a follow up, simply trying a simpler test case. EG (7/11/2001 11:31:14 AM) moving to DUI DS (7/12/01 11:12:44 AM) I am not able to reproduce....1+1 works on Sun 1.3. I am setting up to attempt the batch compile snippet.

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b48f54dc1c25ebdb20fc'}

The Debug menu doesn't work outside of the debug perspective 1. Debug a program. 2. After you're automatically put in the debug perspective, switch to the java perspective. Notice that most of the menu items in the debug menu are grayed out 3. Switch back to the debug perspective and do a step over. 4. Go back to the java perspective Notice that the debug menu is enabled, but when you click any of its menu items, nothing happens NOTES: DS (7/25/01 2:46:13 PM) Moving to ITPDUI. DS (8/7/01 3:13:17 PM) Fixed partially. Waiting on 1GHUHFQ: ITPUI:WIN2000 - Missing notifications in SelectionService.

normal

CLOSED

WONTFIX

{'$oid': '52e9b48f54dc1c25ebdb20fd'}

DS (7/10/01 12:06:44 PM) Correct mechanism for getting a display when not in the UI thread Processing of runnables during/after shutdown/close (disposal of the display or of the control) For example EnabledDisableBreakpointAction which deals with a selection provider...has no way of checking for disposal yet calling #getSelection on the selection provider walksback. Also in the following stack: org.eclipse.swt.widgets.Display.runAsyncMessages (Display.java:1109) at org.eclipse.swt.widgets.Display.readAndDispatch (Display.java:1012) at org.eclipse.swt.widgets.Display.release (Display.java:1062) at org.eclipse.swt.graphics.Device.dispose (Device.java:187) how should the runnables check or know that the display/device is getting disposed. Initializing the settings of a wizard page...see JavaApplicationWizardPage...need to post a runnable such that the initialization (setting focus etc.) always occurs correctly. =override setVisible and do the initialization there...no need for a runnable Actions that are contributed via the view contribution mechanism...setting state in the action and not the delegate on start up and persisting this state (static method filter). 1GDTUNW: ITPDUI:WINNT - Settings in debug views not persisted Instruction pointer marker - how do editors for different languages know to display the marker? - how do editors that are not resource editors (class file) display the marker? - what are the performance concerns of continually updating the line # of the marker, and deleting/creating a new marker as the source we browse changes? 1GDVT53: ITPDUI:ALL - DCR: Highlight current line without using selection marker image provider--extension point --Java guys use this already, Eclipse UI as well The marker annotation models for the resource editors will have to handle the marker similar to breakpoints...if we can configure the instruction pointer to be a marker with attributes for the current source, the icon will just happen using the image provider extension point. SelectionProvider/SelectionService and only provides selection when has focus --> reimplementation in clients that require selection providers that are not focus specific DebugUIPlugin, DebugView, custom wiring of views, etc. created 1GHU8FN: ITPUI:WIN2000 - ISelectionService should allow tracking of specific parts 1GHUGLC: ITPUI:WIN2000 - SelectionService loses source 1GHUHFQ: ITPUI:WIN2000 - Missing notifications in SelectionService Global "relaunch last" action that is contributed to debug menu is always enabled. How can we selectively enable it (i.e. only when there has been a launch, or when there has been a launch of the correct kind - for example we do not want to enable when the last launch was a scrapbook launch). --possible to ignore the first time and then once we are loaded do the right thing. not that great of a solution...better to be consistent? How to specify enabled, disabled, and hover images for contributed action in XML. 1GF64JZ: ITPDUI:WINNT - Rollover icon the same as flat icon @See 1GHUL80: ITPUI:WIN2000 - Contributed actions should have hover and disabled images Global "step" actions (over, into, etc.). These actions are contributed to the debug menu such that we can get hot keys to work. However, the actions are not global - they are specific to a view. We programmatically search for the debug view when an action is invoked. Is there a better way to do this? -improvements in the selection servive to provide the source of the selection and to only register for a specific part will help to address this issue @1GHU8FN: ITPUI:WIN2000 - ISelectionService should allow tracking of specific parts The debugger is intended to support multiple languages. Thus there are different contexts that the UI will need to take one. For example, when debugging java code, the "static & final" filters should be present in the toolbar, but when debugging COBOL code, they do not need to be present. These almost feel like "action sets" for viewers (that are contributed dynamically). Could there be such support? @see 1G9YWCG: ITPDUI:ALL - UI extensions @see 1GHW7P7: ITPUI:ALL - Should have action sets for views as well When a TextViewer (ConsoleViewer) is disposed, it sets its text widget to null...there goes the control so how can check if the thing is disposed? If the control is null, the viewer is disposed. Would be convient for clients if the viewer had #isDisposed: delegated to the control if it is there and returned true otherwise? Other wise we are ending up with the following code: if (fViewer != null) { Control ctrl= fViewer.getControl(); if (ctrl != null && !ctrl.isDisposed()) { ctrl.getDisplay().asyncExec(r); } --do not want to add disposed state to the viewer @see 1GDXD0Y: ITPDUI:ALL - Reduce single editor flicker Setting selection as you are switching to a perspective...review the code of the DebugUIPlugin#switchToDebugPerspective along with the DebugContentProvider#selectAndReveal --probably best to refactory the code so that the suspend select and reveal is taken care of in one place (DebugUIPlugin as may not have the content provider). Rework. @see 1GFDSU7: ITPDUI:ALL - Debugger not properly getting focus could not reproduce waiting on JB for clarification. Short circuiting nested wizards in a lazy manner? 1GCHSGS: ITPPDE:ALL - Finish enablement in Launch dialog is strange 1GCSYAF: ITPDUI:ALL - Setting default launcher with launch element selected 1GE6S2P: ITPJUI:WINNT - Java launcher: Don't prompt for element to launch if there's only one 1GF5XKA: ITPJUI:WIN2000 - First launch offers choice of 1 launcher --attempt to put short circuit logic in the exection action. Minimized or background window notification when interesting 1GD7P7D: ITPJUI:WIN2000 - Debugger doesn't pop to front when breakpoint hit 1GEUZEX: ITPDUI:WIN2000 - Debugger doesn't come to front when breakpoint is hit --no plan for support here. @see 1GDF2VH: ITPJUI:WINNT - run action missing from package view and outliner @see 1GELJC4: ITPDUI:WIN2000 - Console default font @see 1GET0WB: ITPDUI:WINNT - Hot key stepping: hold key down Generalize console with a platform provided output window. We currently have many different consoles - Ant output, Makefile output, and there should be a more general platform mechanism. with NE created 1GHU9B3: ITPUI:WIN2000 - Performance: IWorkbenchPage.findView is slow working with NE we have found places where we are doing unnecessary postings to the display queue. Review. rework findDebugPresentation...code is confusing.

normal

CLOSED

INVALID

{'$oid': '52e9b48f54dc1c25ebdb20fe'}

MA (09.08.2001 11:35:59) When JRE_SRC is not set and you step in to a class from JRE the source attachment wizard comes up. You can change the source attchment variables, but for the JRE variables this is not very helpful, as you would have to go to the 'Installed JRE' page to set the JRE_SRC. I would suggest to not show a wizard when the JAR is from the variable entry JRE_LIB. NOTES: DW (8/29/01 2:39:24 PM) Investigate allowing the user to change the value of the JRE_SRC variable.

minor

RESOLVED

INVALID

{'$oid': '52e9b48f54dc1c25ebdb20ff'}

MA (09.08.2001 11:18:39) 1. assume JRE_LIB points to c:\xxxx\rt.jar. JRE_SRC is left empty 2. create project A that has Variable entry JRE_LIB, Source Attachment set to JRE_SRC 3. create project B that has Library entry to c:\yyyyy\rt.jar 4. create class X in project B with main, System.out.println("Hello World") 5. set breakpoint in B, debug, and step into System. 6. the source attachment wizard comes up, and has filled in JRE_SRC & JRE_SRCROOT -> The SourceAttchmentBlock is fed with a JAR from the project A -> You edit the source attachment for the variable entry. How do you get the classfile to open in the editor? NOTES: DW (8/29/01 2:37:43 PM) The problem here is that the wizard should appear with the context of project B - that is, the rt.jar is c:\yyyyy\rt.jar, and there is no source attachment. DS (9/13/01 11:07:52 AM) I am having trouble reproducing the problem with steps above. The source attachment wizard comes up and is filled with the library entry c:\yyyy\rt.jar. Sent email to MA. MA (27.09.2001 09:49:02) It could be the problem that all entries from JAR files are shared by the JavaCore. So the parent project you get is just any in the workspace. Thats what makes it hard to reproduce. I don't know whats the state of the discussion there, move the PR to JCore?

normal

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2100'}

Test case: Start a self-hosting workbench that was saved with 2 windows opened: - Java perspective - Hierarchy perspective (opened on JavaElement) Running under the debugger with: - Sun's JDK1.3 (build 1.3.0-H): 27min 17s - Sun's JDK1.2.2 (build JDK-1.2.2_005): 3min 13s - IBM's JDK1.3 (build cn130-20010502): 1min 59s - J9 (build 20010510): 3min 21s Running with no debugger: - Sun's JDK1.3 (build 1.3.0-H): 1min 17s - Sun's JDK1.2.2 (build JDK-1.2.2_005): 1min 21s - IBM's JDK1.3 (build cn130-20010502): 1min 22s - J9 (build 20010510): 1min 23s Runing inside VAJ (INTERNAL-DEVELOPMENT - 2.0 (NC) 28/11/2000 [JDK1.2] IV3.5: 1min 38s NOTES: JBL (7/10/2001 4:07:48 PM) There seems to be a bad performance problem with Sun's JDK1.3 (27minutes compared to 2minutes with IBM's JDK1.3). DS (7/10/01 4:53:22 PM) I am not seeing your kind of times for SUN JDK1.3 I have profiled (using OptimizeIt) a self hosting session to bringing up a workbench in debug mode and that only took ~20 minutes. I did not have exactly the workbench you described but... JBL (7/11/2001 10:55:23 AM) Here are the projects I have loaded in my development workbench: Eclipse Core Launcher (binary: startup.jar) org.apache.xerces (binary) org.eclipse.ant.core (binary) org.eclipse.ant.ui (binary) org.eclipse.compare (binary) org.eclipse.core.boot (binary) org.eclipse.core.resources (binary) org.eclipse.core.runtime (binary) org.eclipse.core.target (binary) org.eclipse.debug.core (binary) org.eclipse.debug.ui (binary) org.eclipse.help (binary) org.eclipse.help.ui (binary) org.eclipse.jdt.core (source) org.eclipse.jdt.debug (source) org.eclipse.jdt.doc.isv (binary) org.eclipse.jdt.doc.user (binary) org.eclipse.jdt.launching (source) org.eclipse.jdt.launching.j9 (source) org.eclipse.jdt.ui (source) org.eclipse.jdt.ui.vcm (source) org.eclipse.pde (binary) org.eclipse.pde.core (binary) org.eclipse.pde.doc.user (binary) org.eclipse.pde.runtime (binary) org.eclipse.platform.doc.isv (binary) org.eclipse.platform.doc.user (binary) org.eclipse.scripting (binary) org.eclipse.search (binary) org.eclipse.swt (binary) org.eclipse.ui (binary) org.eclipse.update (binary) org.eclipse.vcm.core (binary) org.eclipse.vcm.core.cvs (binary) org.eclipse.vcm.core.cvs.ssh (binary) org.eclipse.vcm.ui (binary) org.eclipse.vcm.ui.cvs (binary) org.eclipse.webdav (binary) Example of the layout of a source project: org.eclipse.jdt.core - External jars: /org.eclipse.core.resources/resources.jar /org.apache.xerces/xerces.jar /org.eclipse.core.runtime/runtime.jar - Variables: path="JRE_LIB" rootpath="JRE_SRCROOT" sourcepath="JRE_SRC - Source folders: Eclipse Java Batch Compiler Eclipse Java Code Assist Eclipse Java Compiler Eclipse Java Core Search Eclipse Java Evaluation Eclipse Java Model Eclipse Java Formatter Example of the layout of a binary project: org.eclipse.ui - Internal jars: workbench.jar - External jars: /org.apache.xerces/xerces.jar /org.eclipse.core.runtime/runtime.jar /org.eclipse.core.resources/resources.jar /org.eclipse.help/help.jar /org.eclipse.swt/swt.jar - Variables: path="JRE_LIB" rootpath="JRE_SRCROOT" sourcepath="JRE_SRC" They all use the same JRE_LIB which is Sun's JDK1.3 rt.jar. The content of the target workbench is exactly the same (made a copy of the development one). I have the above 4 JREs define in my Preferences->Java->Installed JREs Before each test, I set a custom JRE in the Eclipse Core Launcher project's Properties. I launch the target workbench using UIMain in startup.jar with the following program arguments: -dev bin -data d:\eclipse\target\plugins I start my stopwatch when I press the 'running man' or the 'debug' button, and I stop it when the hierarchy perspective on JavaElement is up. DS (7/18/01 4:16:53 PM) Using JBL's workspace, after ~10 mins I consistently lock up: 1GH2UOQ: ITPCORE:ALL - Deadlock starting workspace in debug mode on Sun 1.3 But we are very busy for those first 10 minutes...can investigate that part at least

normal

CLOSED

INVALID

{'$oid': '52e9b48f54dc1c25ebdb2101'}

DW (7/17/01 7:22:47 PM) When I put a breakpoint on a line like String x = "Hello"; I must press step over twice get over the line. The breakpoint event appears to happen twice. Resume presents the same problem. DW (7/18/01 8:51:32 AM) Not critical. DW (7/18/01 1:10:03 PM) Only occurs with IBM 1.3 VM. Sun VM 1.2.2 works as expected.

normal

CLOSED

INVALID

{'$oid': '52e9b48f54dc1c25ebdb2102'}

The compiler uses char arrays instead of Strings. char arrays don't print nicely. That's why we have these nice toString() methods to help us debugging. The VA/Java debugger used to show them. The Eclipse debugger doesn't which makes debugging the compiler very painful. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2103'}

I cannot get breakpoints to be hit when located on a method declaration and its arguments are layed out on several lines: class X { void foo(String a, String b, <--- breakpoint on this line is not hit String c) { System.out.println(); } } NOTES: DW (7/30/01 9:59:27 AM) Does it ever show as installed? (I.e. change from blue to green?). DS (7/30/01 10:01:12 AM) The breakpoint is not installed. DW (7/30/01 11:38:48 AM) This is a dup of our smarter breakpoint location verification bug.

normal

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2104'}

1. Create CU Test.java in default package: [public class Test { public static void foo() { System.out.println("foo"); System.out.println("breakpoint"); // breakpoint on this line } public static void bar() { foo(); } public static void main(String[] args) throws Exception { Class clazz = Test.class; java.lang.reflect.Method method = clazz.getMethod("bar", new Class[] {}); method.invoke(null, new Object[] {}); } } ] 2. Put a breakpoint in foo() where specified 3. Run under debug with the J9 JRE 4. In the debugger, select the stack frame foo() 5. Drop to frame Observe: the cursor is positioned on the first line of foo() as expected 6. Select the stack frame bar() 7. Drop to frame Observe: the thread is marked as 'Stepping' and collapses after 2 seconds. It never returns from the drop to frame action. NOTES: DS (7/19/01 1:41:56 PM) Created PR for us to change our IJavaStackFrame interface to not claim to stop at breakpoints. Seems to be related to the refection call. For some reason in this case you are hitting the breakpoint in foo...appears like doing a "normal" run to return, instead of the normal behaviour of HCR ThreadReference#doReturn which does not hit breakpoints. Changing main to : public static void main(String[] args) throws Exception { Class clazz = Test.class; bar(); } and things work as expected. Emailed MM. MM (7/25/2001 6:03:36 PM) I traced JDI in the Eclipse debugger with the two versions of the program, one with a DIRECT invocation of bar and one with a 'REFLECT' invocation. I enabled tracing by putting a jdi.ini file in jdimodel.jar and sent both traces to both Darins :-) If you look at the DIRECT trace, the first FRAMES command shows 3 stackframes, all with the same classType corresponding to the testclass, 'test/StackframeTest'. The REFLECT trace shows 4 stackframes, 2 with classType 'test/StackframeTest' on top, below that one with classType "java/lang/reflect/Method" and below that one with classType 'test/StackframeTest'. Both traces show three 'HOT_CODE_REPLACEMENT - DO_RETURN' commands. I think in the REFLECT case there should have been one DO RETURN command more. The STEP event in the REFLECT case just after the last DO RETURN has classType "java/lang/reflect/Method". The STEP event in the DIRECT case just after the last DO RETURN has classType 'test/StackframeTest'. DS (7/26/01 12:59:07 PM) Simplier test case: Put a breakpoint in bar() Hit breakpoint, remove breakpoint Attempt to drop to invoke frame In Eclipse you end up back in main In VAME, the drop to frame action is disabled for both the invoke and main frames. I will attempt to look at the VAME debugger source and talk to JBL. DS (7/30/01 10:31:30 AM) This is the code that appears in the VAME debugger... // Cannot drop if the current frame is native, or there is a native above it int currentIndex= fFrames.indexOfElement(frame); for (int i= 0; i <= currentIndex; i++) { DbgStackFrame f= (DbgStackFrame)fFrames.elementAt(i); if (f.isNative()) { return false; } } DS (7/30/01 11:05:26 AM) java.lang.reflect.Method#invoke returns true for #isNative() Need to find out from the VAME debugger implementers where this restriction came from. Sent email. DS (7/31/01 9:57:33 AM) @see 1GHQHUB: ITPDUI:ALL - Drop to Frame limited with native methods DW (8/7/01 2:21:37 PM) Fixed. DS (8/7/01 2:57:00 PM) The original test case still does not work for me.

minor

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2105'}

DW (7/26/01 2:23:19 PM) Use of the shared memory transport may give improved debugger performance when debugging locally. Results on WIN98 - not too promising. IBM1.2.2 - VM GPF'd when speficied dt_shmem JDK1.2.2 - Failed to init transport - FATAL ERROR in native method. JDK1.3.0 - D:\workspaces\ws1\Test>\jdk1.3\bin\java -cp . -Xdebug -Xnoagent -Djava.compiler= NONE -Xrunjdwp:transport=dt_shmem,address=8000 Test Java HotSpot(TM) Client VM warning: Setting of property "java.compiler" is ignor ed Error accessing shared memory, rc = -1 Transport dt_shmem failed to initialize, rc = -1. FATAL ERROR in native method: No transports initialized JDK 1.4 D:\workspaces\ws1\Test>\jdk1.4\bin\java -cp . -Xdebug -Xnoagent -Xrunjdwp:transp ort=dt_shmem,address=8000 Test Transport dt_shmem failed to initialize, rc = 509. FATAL ERROR in native method: No transports initialized JGS (8/2/01 6:23:41 PM) The dt_shmem transport seems to work fine on 98, NT and 2000 on JDK1.2.2 & JDK1.3.0. JDK1.4 fails, but may be because of misplaced JDI DLLs. JGS (8/6/01 9:53:49 AM) The JDI DLLs are misplaced in the JDK1.4 beta. They ship in \jre\bin but should be moved to \bin. When this is done, dt_shmem works fine. JGS (8/6/01 2:16:40 PM) JDI Implementation does not currently support shared memory transport, so we cannot play with this in Eclipse or do timings to compare to socket transport.

enhancement

RESOLVED

WONTFIX

{'$oid': '52e9b48f54dc1c25ebdb2106'}

DS (9/7/2001 10:46:25 AM) From RGP I just had to debug a deadlock and one thing that would be nice is to show a little "synchronized" icon in the stack trace to indicate synchronized methods and possibly synchronized blocks. NOTES:

enhancement

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2107'}

DS (7/30/01 1:29:40 PM) Using the following source =================== public class InfiniteLoop { public static void main(String[] args) { (new InfiniteLoop()).loop(0); } public void loop(int i) { while (true) { System.out.println("Looping " + i); i++; try { Thread.sleep(1000); } catch (InterruptedException e) { } } } } public class InfiniteLoop { public static void main(String[] args) { (new InfiniteLoop()).loop(0); } public void loop(int i) { while (true) { System.out.println("Looping " + i); i++; try { Thread.sleep(1000); } catch (InterruptedException e) { } } } } =================== put breakpoints on both method declaration lines. Only the breakpoint for main shows as installed. NOTES:

normal

RESOLVED

WORKSFORME

{'$oid': '52e9b48f54dc1c25ebdb2108'}

98% of the time I am not interested in terminated processes remaining in my debug view. I would like an option which clears the view of terminated processes when a new process is added. Thus I would always have the latest process to see console output but it would not become cluttered with dead processes. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb2109'}

The variable values don't display nicely when then are on several lines. The Eclipse debugger should use a text widget (as the VAJ debugger used to) to display the value of a variable. NOTES: DW (7/23/01 9:54:27 AM) Should talk with JJ to determine if varaible UI could be changed to be more like VAJ. This would allow more efficient "toString" implementation and better presentation of variables that take up > 1 line. Also would provide a UI for doing evaluations withing the the result of an evaluation. DW (7/24/01 9:36:04 AM) Sent email to JJ and EG re variables view presentation.

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b48f54dc1c25ebdb210a'}

I apply some changes in order to be able to evaluate inside binaries. It works fine for classes inside a jar which is not sealed. If the jar is sealed, I get a java.lang.SecurityException. See 1GE4ILR: ITPJCORE:ALL - Eval - Evaluation in Binary Project fails (02/08/2001 11:09:59 AM) NOTES: DW (8/8/01 9:52:20 AM) Not sure if there is anything we can do about this.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b48f54dc1c25ebdb210b'}

When a method frame is selected in the debugger, the variables view should always include the parameter values. This information is crucial to see what values a method is being passed, and is ESPECIALLY CRUCIAL - when source code for the method is not available - when debug info for the method is not available Currently, the debugger shows the value of the pseudo variable "this" but only includes the values of the parameter variable when source code is available for the method. When the names of parameter variables cannot be determined from the class file debug info or source code, use meaningful invented names like "arg1", "arg2", etc. so that the user will immediately understand what's going on. NOTES: DW (8/7/01 1:28:01 PM) JDI will not allow us to retrieve parameter values unless debug info is present. It is not dependent on source code. @see javadoc for com.sun.jdi.StackFrame.visibleVariables(). In order to retrieve the value of a variable, we require a LocalVariable object. If these are not returned from #visibleVariables, we are unable to show any info. jeem (8/7/01 3:03:01 PM) This is an unfortunate limitation of JDI. I recommend submitting a PR to Sun for this one. JDPA could provide argument LocalVariable objects for parameter variables even when there is no debug info for the method. The number and types of the parameter variables is given by the type info in the method header; the offsets of the parameter variables is predictable and can be readily calculated from info in the method header. DW (8/10/01 1:33:21 PM) Bug submitted to Sun: ************************************************ Your report has been assigned an internal review ID of: 129819 This review ID is NOT visible on the "Java Developer Connection" (JDC). We greatly appreciate your interest in improving the quality of Java(tm) Technology from Sun Microsystems. Please be aware that the large volume of reports we receive sometimes prevents us from responding individually to each message. We currently have a three week response time for responding to Bug Reports. If the information is determined to be a new bug, or a duplicate of a known bug, you will receive a followup email containing a seven digit bug number. You may search for this bug number on the "Java Developer Connection" (JDC) at this URL: http://developer.java.sun.com/developer/bugParade/index.html. If you just reported an issue that could have a major impact on your project and you require a response, please consider purchasing one of the support offerings at this URL: http://java.sun.com/support/index.html -------------------------------------------- dateCreated: Fri Aug 10 12:54:33 MDT 2001 type: rfe cust_name: Darin Wright cust_email: Darin_Wright@oti.com jdcid: status: Waiting category: java subcategory: debugger company: Object Technology International release: 1.3 hardware: generic OSversion: generic priority: 4 synopsis: JDI - provide arguments when no debug attributes present description: java version "1.3.0" Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.0-C) Java HotSpot(TM) Client VM (build 1.3.0-C, mixed mode) Request for argument LocalVariables when no debug attributes present. Currently, the JDI specification and implementaion of com.sun.jdi.StackFrame.visibleVariables() throws an AbsentInformationException if there is no line number information for a method. However, it would seem that it should be possible to return LocalVariable objects for the method arguments in this case, as the number and types of the parameter variables is given by the type info in the method header; the offsets of the parameter variables is predictable and can be readily calculated from info in the method header. Variable names may have to be invented - i.e. "arg1", etc. It would be advantageous to display arguments & their values in a debugger even when debug attributes are not present. workaround: comments: (company - Object Technology International , email - Darin_Wright@oti.com) DW (8/10/01 3:43:05 PM) The report you submitted has been determined to be a new RFE. It has been entered into our internal bug tracking system with the assigned Bug Id: 4490824. The state of the bug can be monitored via the The Java Developer Connection Bug Parade at: http://developer.java.sun.com/developer/bugParade/index.jshtml It may take a day or two before your bug shows up in this external database. The Java Developer Connection is a free channel that is maintained by staff here at Sun. Access this web page to join: http://developer.java.sun.com/servlet/RegistrationServlet The home page for the Java Developer Connection is: http://java.sun.com/jdc

enhancement

VERIFIED

FIXED

{'$oid': '52e9b48f54dc1c25ebdb210c'}

Running Eclipse in Eclipse scenario, using latest jdtcore. Placed a breakpoint in JavaModelManager.readOptions(...) (only existing in latest 2.0 stream). Then hit the breakpoint by adding a new Java project in empty workspace. Tried to evaluate expression 'fOptions' (first occurrence in the method). The debugger seemed to have resumed. Breakpoint stack frames where gone, could not interact other than killing threads. When closing debugger, the entire IDE disappeared. NOTES: DW (8/29/01 3:14:12 PM) Try to reproduce.

minor

RESOLVED

INVALID

{'$oid': '52e9b48f54dc1c25ebdb210d'}

DS (8/3/2001 10:20:43 AM) Is the plan to keep the layout perspective id supplied from the launcher? NOTES: DW (8/7/01 9:02:05 AM) Yet unsure. We have also noticed the need for some support for a debug "context" to know which views and buttons to display.

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b49054dc1c25ebdb210e'}

We need to add a mechanism to the JDIThread to perform "quiet" suspends/resumes which do not update the UI. This functionality would be useful for "normal" (user initiated) evaluations, scrapbook evaluations, and for conditional breakpoint evaluation. In all three cases, we are doing "artificial" suspensions (to perform evaluations) that the user should not know about. In the case of normal evaluation and conditional breakpoint evaluation, the lack of a quiet suspension mechanism means that the UI flashes rapidly as the JDIThread is suspended and resumed repeatedly. When the user runs a scrapbook page in the Debug Perspective, the debug view shows too much of the underlying implementation of the scrapbook. After the user runs code, the debug view shows them that the underlying thread is suspended. This is confusing to the user because they're being presented with the option to resume but they have no idea why anything would be suspended. Something to consider is the fact that if a user-defined breakpoint is hit while evaluating code, we'll have to have a way to override the quiet-mode of the JDIThread to update the UI appropriately. This can be accomplished (I've implemented it once minus the "something to consider") by: 1) Add a quiet flag to the JDIThread. 2) Add suspendQuiet which sets the quiet flag true and calls an internal suspension method 3) Add a check in the internal suspension and old resume methods. If the quiet flag is true, don't update the UI. 4) The old resume method should turn quiet mode off. The old suspend method should turn it on and call the internal suspension method. 5) Add public method to turn quiet mode off and another to fire an explicit UI update. A) When a conditional breakpoint is hit, call the suspendQuiet method, and perform the evaluation, if the evaluation comes back true, tell the thread to notify the UI of suspension. If not, resume execution. B) When the scrapbook is finished with an evaluation, it should call the suspendQuiet method. C) When a breakpoint without a condition is hit, suspend as usual (call the old suspend method). Calling the old suspend method turns quiet mode off (see step 4) so the UI will update as usual for breakpoints hit when the thread is in quiet mode (during conditional breakpoint or scrapbook evaluation). NOTES:

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b49054dc1c25ebdb210f'}

DS (8/10/01 9:53:33 AM) <<Using the InfiniteLoop program, with the sleep bumped up to 4000>> Set a breakpoint with a hitcount of 3 on the System.out line. Once 2 iterations have occurred (to System.outs to the console), reset the hitcount to 4 4 more iterations of the loop will occur until the breakpoint is hit. Reset the hitcount to 3 on the expired breakpoint Once 2 iterations have occurred (to System.outs to the console), reset the hitcount to 3 Only 1 more iteration will occur. Seems we are not resetting the count if the count is the same. NOTES: JMB (8/13/2001 10:55:31 AM) Defer until IBreakpoint refactoring. In the IBreakpoint universe the fix is for LineBreakpoint>>setHitCount to generate a new request everytime it is called (if the breakpoint is installed). JMB (9/12/2001 9:21:23 AM) This is more complicated than I first suspected, since breakpoints keep no record of which targets they're installed in. When a target calls IJavaBreakpoint#changeForTarget(this), the breakpoint must somehow figure out if the hitCount needs to be updated. Simply comparing the marker hit count to the request hit count doesn't solve the problem of setting the hit count to the same value. The fix may require the creation of a new breakpoint change mechanism whereby we pass a code which tells us which attributes have been updated.

minor

RESOLVED

WORKSFORME

{'$oid': '52e9b49054dc1c25ebdb2110'}

DW (8/10/01 11:00:27 AM) When a breakpoint is hit while doing a step, the step event may be received after the breakpoint. The breakpoint has already ended (aborted) the step. It looks like our model does more work than necessary to process the step event.

minor

RESOLVED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb2111'}

DW (8/10/01 11:50:29 AM) The breakpoint view could refelect the selected target/session. For example, breakpoints that are installed, can show installation status, based on the selected target, since a breakpoint not need be installed in all targets.

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b49054dc1c25ebdb2112'}

DW (8/10/01 1:26:59 PM) When running the JDIModelPresentationTests, without first manually openning the Debug and Java perspectives, a deadlock occurrs. When the test is run without the Debug UI perspective already loaded, it causes the plugin to startup, which registeres all active launches. Registering of a launch is handled with a sync exec in the debug UI, which hangs forever.

minor

RESOLVED

WORKSFORME

{'$oid': '52e9b49054dc1c25ebdb2113'}

DS (8/14/01 11:34:18 AM) I had inspected an instance of DebugPlugin, with the intent of looking at event listeners being added and removed. fEventListeners.fListeners did not always reflect the current state Inspecting fEventListeners.fListeners gave an inspect item with the correct state. To reproduce: Debug a self host put a breakpoint in DebugPlugin#addDebugEventListener in the self host debug a program when the breakpoint is hit, inspect DebugPlugin repeat launching (f10) and notice that the listener list does not always grow with an addition. NOTES:

minor

VERIFIED

WORKSFORME

{'$oid': '52e9b49054dc1c25ebdb2114'}

DW (8/14/01 10:33:38 AM) The Java Debug UI, launching, and Snippet Editor should be in a seperate plugin.

normal

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb2115'}

DW (8/14/01 4:05:37 PM) Need to identify the places where we rely on resources, java model, and java UI support.

enhancement

CLOSED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb2116'}

DS (8/14/01 3:45:54 PM) We have a few actions (CopyToClipboardActionDelegate...) that are delegates that don't need to be. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb2117'}

DS (8/15/01 2:08:46 PM) Wouldn't it be better not show an editor at all for .class files that have no source? Is it possible to drive the outline without the editor...it seems to add nothing of value. NOTES:

normal

RESOLVED

WORKSFORME

{'$oid': '52e9b49054dc1c25ebdb2118'}

JGS (9/10/01 3:48:09 PM) We should consider moving some of the actions that currently reside on the toolbars of our views to the menu that drops down from the triangle icon on the toolbar. For example, we could move the "Show (qualified, final, etc.)" type actions. This would be in keeping with the types of actions that the workbench puts on this menu ("Show version info"), and could potentially save some toolbar bar real-estate. In particular, we should also consider if the "ToggleStepFilterAction" should be moved to the drop-down menu for the DebugView. CM (9/12/2001 12:20:51 PM) These kind of actions are always quick. Click-click. i.e. the thought process is: "What class is that variable anyhow?" ...click tool... "Vector... Now I want my variables tree to be lean again" ...click tool again to make the extra information go away... Note that the toolbar wraps if the view is too small.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb2119'}

Ideally there should be a way to indicate a scope which is used to stop if the exception is uncaught in this given scope. Typically, for Eclipse-in-Eclipse, I am interested in exception uncaught in the jdtcore, knowing that they will eventually get caught by the platform. Similarily the JUnit tests are catching exceptions at some point. I am wondering if the notion of search scope could not be reused in that sense... I would be using my user defined scope for more than one purpose, which would make them more appealing. NOTES: DS (7/30/01 2:36:11 PM) Another interesting idea here (out of our step filtering idea), is to not worry about exceptions that come out of class libraries (do not show exceptions thrown out of these packages/classes). DW (8/28/01 9:57:29 AM) Also a request to allow exclusion filter: "In the debugger I had a BP for exceptions. However the code I am dealing with throws this exception a lot, in a particular function where I don't have control over. It would be handy to have an exclusion list available where the exception would be ignored."

normal

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb211a'}

Currently, we have buttons to toggle "Show Qualified Names" in three windows (Display, Inspect, Debug) and "Show Type Names" in two (Display, Inspect). Instead of having to set these preferences in each window, suggest that it might be more convenient for them to be global preferences with the primitive type display options (on the new JavaDebugPreferencesPage). NOTES:

normal

RESOLVED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb211b'}

JGS (8/16/01 4:06:34 PM) We should investigate putting compile errors that results from a scrapbook evaluation into the task list. At present, they are inserted into the scrapbook document just before the snippet.

normal

CLOSED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb211c'}

JUnit setup 1) set a breakpoint in junit.samples.VectorTest.setUp() 2) run VectorTest 3) modify setUp method and rebuild -> the debugger drops to the method instead of to the entry of setUp. public static void main (String[] args) { junit.textui.TestRunner.run (suite()); } NOTES: JMB (8/20/2001 11:20:22 AM) Duplicated with simple test case (main() calls method foo() in any class). The problem is that we receive notification of CompilationUnit changes, not method changes. Since we don't know exactly which method was altered, dropping to the deepest method in the stack for the altered CompilationUnit is the safest thing to do. It is unfortunate, but given the current limitations, the current behavior seems to be the best we can do. TM (8/21/01 10:54:41 AM) Could we do a method delta locally? We could get the old class file from the local history. If it's not there anymore, we could always revert to the current behaviour. JMB (8/21/2001 10:58:12 AM) An alternate suggestion (from Erich) is that we change the default behavior to drop to the top stack frame (instead of the bottom) and warn the user if there are more stack frames below, possibly giving them the option to drop more. Everyone's intuition seems to be that the common case is to drop to the top frame, so this may be the way to go. EG (22.08.2001 11:49:57) The class files are not kept in the local history. We didn't want to bloat the local history with class files. EG (04.09.2001 17:16:44) Another idea, we should be able to find out when a user modifies a method in the the context of the debugger. Drop to frame would then only drop to the modified method. DW (9/17/01 3:03:41 PM) Listen to JavaModel deltas instead of resource deltas. When auto-build is on, we notify the target of updates. Use the detailed deltas to determine where to drop to (and if the HCR should be done - i.e. ignore when there are shape changes). JMB (9/20/2001 10:17:39 AM) After further exploration it turns out that JavaElementDeltas don't give us any fine grained information (method changes, etc.). DW (9/24/2001 8:30:21 AM) Fine grained info is present, but the deltas only provide add/remove info, not "changed' info for members, which would be required to determine which stack frames to drop.

normal

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb211d'}

DW (8/29/01 1:20:26 PM) From org.eclipse "When I get an error, I get a trace, it would be good to be able to select the Exception class, right click and add a breakpoint and re-run the execution up to the exception..."

enhancement

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb211e'}

Stepping through org.eclipse.swt.layout.GridLayout.calculateGridDimensions(...), I found that the variables it declared in the first statement as: int maxWidth, childWidth, maxHeight, childHeight; would appear and disappear in the variables view as I stepped through the code They did not appear until first assigned. childWidth disappeared when I exited the nested for loop for row. maxWidth disappeared when I exited the for loop for column. I would expect them to always be visible after the first statement, or at least after they're first assigned, until the method returns. NOTES: NE (08/29/01 6:04:59 PM) Screen shot in Q:\viper\team\Nick\PRs\1GJ8PX4.zip DW (9/7/01 8:56:26 AM) I think this is due to the way the byte codes are generated. The variables are allocated inside the loops, even though they are declared outside the loops. Waiting for clarification from PM.

normal

RESOLVED

INVALID

{'$oid': '52e9b49054dc1c25ebdb211f'}

MA (22.08.2001 18:32:36) The debug and run button on the toolbar contains a very helpful history. I changed the program arguments and selected the history entry, but the new program arguments were never taken. I had to select the elemnt and choose debug>Java Application. I guess it's a relaunch, but I was expecting it to just be a store of selection and used launcher. Program arguments would be then be read again. NOTES: DW (8/29/01 2:25:38 PM) New program arguments should be used. Not sure why this is not working, as a launch history element is just an element & launcher pair. DS (9/14/01 3:14:08 PM) Cannot reproduce on build 135. Test case Run Args <<output>> Testing program arguments Set the program arguments to "testing" Using the Run history, run args <<output>> testing Testing program arguments Source for Args public class Args { public static void main(String[] args) { if (args.length > 0) { System.out.println(args[0]); } System.out.println("Testing program arguments"); } } Sent email to MA. DS (9/17/01 3:23:36 PM) From MA Currently JavaModel shares the elements in a JAR in a workspace. So for JARs that are in two (or more) projects, the getJavaProject does return you any of these projects, not necessary the one you found it in. (PM is saying that getJavaProject should not be used for JAR elements). I am always launching from elements in JAR (UIMain in startup.jar), and after having the test projects in the workspace (containing startup.jar as well), I got strange effects, classpath for launching not correct set up, changing did not help, arguments vanishing... There is nothing you and I can do, I guess, I think PM has to remove the sharing. EG is in discussion with him.

minor

RESOLVED

WONTFIX

{'$oid': '52e9b49054dc1c25ebdb2120'}

This request comes from the Eclipse Corner newsgroup: It would be nice to be able to supply a formatter for inspection of Objects. For example, a Hashtable is typically hard to read in the Inspector. Sometimes it is sparsely populated, sometimes Entries are linked with a next pointer, but a formatter could take that input Tree, and return an Output Tree that looks like this. +Hashtable(size = 21) |- - Entry ("key 1") | |- Object (@a32) |- + Entry("key 3") |- + Entry("key 0") This could be done without running code on the debugging VM. But even more powerful would be the ability to run code on that VM, then one could do something like: Composite |- - children | | - Button | | - Button |- flags Note that children is not even a field on Composite, but can be obtained by calling getCompents(). I assume this might be possible since Micro Edition would call toString() on the debug VM, but this may be part of the Debug spec. NOTES:

enhancement

VERIFIED

FIXED

{'$oid': '52e9b49054dc1c25ebdb2121'}

We should support conditional breakpoints. The concept seems valid for line breakpoints, method entry breakpoints, and watchpoints. NOTES: JMB (8/24/2001 3:42:38 PM) I've prototyped this and it was easy to get conditional breakpoints working for simple cases (simple variable equality checking and function calls). In the case of conditions which fail evaluation (because of user error *or* failure of our evaluation engine), we should suspend for the breakpoint, pop up a dialog informing the user that the evaluation of their condition failed, and disable the breakpoint.

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2122'}

EG (24.08.2001 12:10:42) This is a request for another flavor of conditional breakpoints. We should support that a user can constrain a breakpoint to a particular process or thread. This is useful for multi-thread/multi-process debugging when both are executing the same code. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2123'}

EG (24.08.2001 12:12:57) there can be many different launchers, e.g., in the WSAD case. As a consequence the menu can get large and it is tedious to select a particular launcher. We should investigate whether we support to filter the set of available launchers per perspective. For example, in the Java perspective I don't need to have access to the C launcher. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2124'}

EG (24.08.2001 12:16:16) low prio feature for Inspector. Provide a field to enter an expression with support for an expression history. NOTES:

enhancement

CLOSED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb2125'}

0. Self hosting setup 1. Launch UIMain (via running man icon) program args: -dev bin -data d:\eoe vm args: -mx350M -ms:20 2. Close the launched workbench 3. Press UIMain from the launch history: UIMain [Java Application] ==> does not work (parameters are not stored) Can be tested by looking at the properties of the two processes: the first one contains the program args for UIMain while the second does not. NOTES: JGS (8/24/01 3:30:58 PM) Cannot duplicate. Sent email to DM for more details. DM (8/27/01 9:14:14 AM) Same in my workspace: HelloWorld works BUT if you follow my test case above i.e launch Eclipse's (out of the 2.0 stream) UIMain then it does not. It does not work either with SDK nor J9. main properties for the first launch: C:\Java\J9_05112001\ive\bin\j9 -classpath d:\EclipseWorkspace\plugins\Eclipse Core Launcher;d:\EclipseWorkspace\plugins\Eclipse Core Launcher\startup.jar;d:\EclipseWorkspace\plugins\org.apache.xerces;d:\EclipseWorkspace\plugins\org.apache.xerces\xerces.jar;d:\EclipseWorkspace\plugins\org.eclipse.ant.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.ant.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.compare\bin;d:\EclipseWorkspace\plugins\org.eclipse.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.swt;d:\EclipseWorkspace\plugins\org.eclipse.swt\swt.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.runtime\runtime.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.resources\resources.jar;d:\EclipseWorkspace\plugins\org.eclipse.help\help.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.boot\boot.jar;C:\Java\J9_05112001\ive\lib\jclMax\classes.zip;d:\EclipseWorkspace\plugins\org.eclipse.core.boot;d:\EclipseWorkspace\plugins\org.eclipse.core.resources;d:\EclipseWorkspace\plugins\org.eclipse.core.runtime;d:\EclipseWorkspace\plugins\org.eclipse.core.target\bin;d:\EclipseWorkspace\plugins\org.eclipse.debug.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.debug.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.help;d:\EclipseWorkspace\plugins\org.eclipse.help.ui;d:\EclipseWorkspace\plugins\org.eclipse.help.ui\helpworkbench.jar;d:\EclipseWorkspace\plugins\org.eclipse.jdt.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.debug\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.debug\jdi.jar;d:\EclipseWorkspace\plugins\org.eclipse.jdt.launching\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.launching.j9\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.search\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.ui.vcm\bin;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core\vcm.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.ui\vcmui.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core.cvs.ssh;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core.cvs.ssh\vcmcvsssh.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.ui -mx350M -ms:20 org.eclipse.core.launcher.UIMain -dev bin -data d:\eoe main properties for the second launch (via menu) ==> no args C:\Java\J9_05112001\ive\bin\j9 -classpath d:\EclipseWorkspace\plugins\Eclipse Core Launcher;d:\EclipseWorkspace\plugins\Eclipse Core Launcher\startup.jar;d:\EclipseWorkspace\plugins\org.apache.xerces;d:\EclipseWorkspace\plugins\org.apache.xerces\xerces.jar;d:\EclipseWorkspace\plugins\org.eclipse.ant.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.ant.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.compare\bin;d:\EclipseWorkspace\plugins\org.eclipse.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.swt;d:\EclipseWorkspace\plugins\org.eclipse.swt\swt.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.runtime\runtime.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.resources\resources.jar;d:\EclipseWorkspace\plugins\org.eclipse.help\help.jar;d:\EclipseWorkspace\plugins\org.eclipse.core.boot\boot.jar;C:\Java\J9_05112001\ive\lib\jclMax\classes.zip;d:\EclipseWorkspace\plugins\org.eclipse.core.boot;d:\EclipseWorkspace\plugins\org.eclipse.core.resources;d:\EclipseWorkspace\plugins\org.eclipse.core.runtime;d:\EclipseWorkspace\plugins\org.eclipse.core.target\bin;d:\EclipseWorkspace\plugins\org.eclipse.debug.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.debug.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.help;d:\EclipseWorkspace\plugins\org.eclipse.help.ui;d:\EclipseWorkspace\plugins\org.eclipse.help.ui\helpworkbench.jar;d:\EclipseWorkspace\plugins\org.eclipse.jdt.core\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.debug\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.debug\jdi.jar;d:\EclipseWorkspace\plugins\org.eclipse.jdt.launching\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.launching.j9\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.ui\bin;d:\EclipseWorkspace\plugins\org.eclipse.search\bin;d:\EclipseWorkspace\plugins\org.eclipse.jdt.ui.vcm\bin;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core\vcm.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.ui\vcmui.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core.cvs.ssh;d:\EclipseWorkspace\plugins\org.eclipse.vcm.core.cvs.ssh\vcmcvsssh.jar;d:\EclipseWorkspace\plugins\org.eclipse.vcm.ui org.eclipse.core.launcher.UIMain DM (8/27/01 6:09:23 PM) Here's the detailed test case: 1. Extract official 0.132 to a new location 2. Replace the Debug and JDT plug-ins (in eclipse/plug-ins) Note: You have to create them first from the currently released 2.0 dev stream. Or: you use 0.132 as is but then you have start Eclipse on Eclipse on Eclipse to see the bug 3. Start Eclipse 4. Add Eclipse code to the new and empty workspace (self-hosting setup) 5. Add the program arguments to UIMain 6. Launch UIMain (i.e. Eclipse on Eclipse) 7. Close Eclipse (started on Eclipse) 8. Re-launch UIMain via running man menu: UIMain [Java Application] I assume this has to do with the introduction of the launch memento.

critical

RESOLVED

DUPLICATE

{'$oid': '52e9b49154dc1c25ebdb2126'}

Prompted by comments on Eclipse the following behavior is suggested for the Display window: 1) We should go to the next line after displaying a result 2) If the user chooses to Display (via the button or otherwise) and they have no text selected, we should highlight the entire line that the cursor is on and display that. NOTES: JMB (8/30/2001 11:38:52 AM) I think we should also switch the shortcut keys around. In the menu, the 'd' in display and the 'i' in inspect should be the underlined shortcut letters. Also, ctrl+d should display and ctrl+i should inspect. This would not only be a usability improvement, it would also make the display view evaluation consistent with the snippet editor (which already supports ctrl+d and ctrl+i). DS (9/5/01 11:49:11 AM) Shortcut keys and menu accelerators are getting confused here. In a popup menu, the shortcuts are not shown (ctrl+d) but the accelerators are (underlined i). The snippet editor does have duplicate accelerators (n and i). Logged 1GJLAL7: ITPDUI:ALL - Duplicate accelerators in snippet editor context menu As well, inspect is ctrl+q in the snippet editor (ctrl+i is for indent). DS (9/5/01 4:31:23 PM) Having trouble getting the shortcuts (CTRL+D and CTRL+Q) to work...as well as finding shortcuts for display and run that do not clash with other shortcuts. DS (9/7/01 1:51:11 PM) Shortcuts on hold. DS (9/12/01 9:07:28 AM) Added the "select line cursor is on support" to the snippet editor. Looking at ways to add the feature to the display view. DS (9/20/2001 3:01:14 PM) New work for 2.0: support code assist more friendly formatting persisting content DS (9/25/01 2:11:39 PM) Ensure to reuse code between the snippet editor, java editor and display view evaluation support. DS (9/28/2001 2:05:00 PM) First attempt complete. DS (10/2/2001 2:16:53 PM) DisplayView content assist does not present the extra context information after a successful code assist. For example: code assist of "eq" should show up equals(Object) select this choice In the JavaEditor you will get a hover help saying object. We do not get this in the DisplayView. DS (10/2/2001 3:32:32 PM) Fixed. Talk to DW about auto-selecting the line.

enhancement

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2127'}

DW (8/29/01 3:23:12 PM) Enablement has been made for JSP/pattern breakpoints. We will also need this functionallity in the 2.0 stream.

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2128'}

When I closed the target Eclipse, the host Eclipse window was not activated, even though it was the next window in the order. This prevented me from simply hitting F10 again to relaunch. NOTES: DW (8/30/01 11:19:09 AM) Investigate - but I suspect this is not a debugger issue.

minor

VERIFIED

WORKSFORME

{'$oid': '52e9b49154dc1c25ebdb2129'}

DW (9/3/01 12:14:23 PM) From eclipse corner: > Some other debuggers will retain expanded variables if a resume is performed > and the expanded objects are again avilable when a break point is > encountered. This is very useful when stoping on each iteration of a loop > and looking for specific values or patterns of values.

enhancement

RESOLVED

DUPLICATE

{'$oid': '52e9b49154dc1c25ebdb212a'}

JGS (9/7/01 9:42:36 AM) JDI does not appear to support this, so we'd need to fake it on the client. JGS (9/10/01 11:28:58 AM) What happens when we step into a synthetic method? Does the debugger keep performing 'step intos' until we get to a non-synthetic method? Similar question for 'step return', does the debugger keep performing 'step returns' until no longer in a synthetic method? And also step overs, since you could step over out of a method and back into a synthetic one. JGS (9/13/01 2:44:21 PM) Will investigate how regular step filtering handles the following situation: - Method 'a' in class 'A' calls method 'b' in class 'B' - Method 'b' in class 'B' calls method 'c' in class 'C' - Set breakpoint in 'a' on call to 'b' - Set step filters such that only B is filtered - When suspended at breakpoint, step in, what happens? - 2 choices: (1) Acts like a step over, or (2) we suspend in 'c' Results of this experiment should guide us on how to handle filtering synthetic methods. If answer is (1), this is relatively straightforward to implement in client. If it's (2), this is much more involved. JGS (9/13/01 3:45:45 PM) It appears the answer is (2). DW (9/24/2001 3:59:05 PM) I think the logic is something like this: * User does a stepX (over/into/return) * If the resulting suspend location is a synthetic method, and the filter is on do another stepX This will get the desired behavior of moving from A to B, and evenutally stepping into C. I also manually insured that stepping "into" will eventually return to B (i.e. if there is nothing to "step into", the result is a "step over"). JGS (10/2/01 11:18:12 AM) Also want to add ability to filter stepping into static initializers. This should be similar to filtering synthetic methods. JGS (10/2/01 11:26:08 AM) Maybe also consider option to filter constructors. JGS (10/2/01 1:10:36 PM) Wait until after refactoring of step handling in JDIThread.

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb212b'}

JGS (9/7/01 9:41:55 AM) Add tests to debug test suite that exercise step filters.

normal

CLOSED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb212c'}

TM (9/7/01 12:28:01 PM) Note: this occurs on WINNT multiprocessor boxes. The JDI Test suite hangs on the VirtualMachineDisposeTest. A little printf'ing show that the test is waiting to connect to the VM. The problem seems to be twofold: 1) The Test should not try to connect to the VM, since the previous tests restore the connection before they finish. 1) No timeout in the testsuite when connecting to the VM. TM (9/14/01 2:47:39 PM) the above is not true. Further analysis shows that the hang actually occurs when doing the handshake with virtual machine when testing the reconnect after VirtualMachine.dispose(). The JDI problem we have here is that the handshake doesn't timeout (SocketTransportImpl.PerformHandshake()). TM (9/14/01 5:06:35 PM) Here's a TestCase: public class HandshakeTimeoutTest extends TestCase { public static final int PORT= 9090; class HandshakeReader implements Runnable { public void run() { try { ServerSocket serverSocket= new ServerSocket(PORT); while (true) { Socket socket= serverSocket.accept(); } } catch (IOException e) { } } }; /** * Constructor for HandshakeTimeoutTest */ public HandshakeTimeoutTest(String name) { super(name); } public static void main(String[] args) { new TestRunner().run(HandshakeTimeoutTest.class); } public void testHandshakeTimeout() { Thread t= new Thread(new Runnable() { public void run() { doTestHandshakeTimeout(); } }); t.start(); try { t.join(10000); assert("Still hanging after 10 seconds", !t.isAlive()); t.stop(); } catch (InterruptedException e) { } } public void doTestHandshakeTimeout() { Thread t= new Thread(new HandshakeReader()); t.setDaemon(true); t.setName("Handshake reader"); t.start(); try { VirtualMachineManager manager = Bootstrap.virtualMachineManager(); List connectors = manager.attachingConnectors(); assert("connectors.size() == 0", connectors.size() != 0); AttachingConnector connector = (AttachingConnector)connectors.get(0); Map args = connector.defaultArguments(); ((Connector.Argument)args.get("port")).setValue(String.valueOf(PORT)); ((Connector.Argument)args.get("hostname")).setValue("localhost"); try { VirtualMachine vm= connector.attach(args); } catch (IllegalConnectorArgumentsException e) { assert(e.toString(), false); } catch (IOException e) { assert(e.toString(), false); } } finally { t.stop(); } } } TM (9/14/01 5:06:57 PM) fixed the JDITestSuite locally to always clean up it's processes. TM (9/28/01 10:39:31 AM) Fixed the proxy and VM to not hang anymore when executing the Test Suite (there was a real bug in that code). Still, we should timeout on the handshake, because obviously, the other side is sick.

normal

CLOSED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb212d'}

DS (9/7/01 2:44:02 PM) Debug to a breakpoint on J9 Edit and save the class you are currently suspended within from the Java perspective You are automatically switched to the debug perspective. I understand why this is happening...just is it the behavior that the user expects? NOTES: DW (9/17/01 11:36:23 AM) I believe this is the correct behaviour - it lets the user know they modified a class that was currently executing/suspended in the debugger.

minor

RESOLVED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb212e'}

DS (9/7/01 2:58:38 PM) Select some code in the snippet editor that has an error and attempt to run the code Hit the backspace key to delete the error message Notice that the state of the run, inspect and display toolbar actions does not update to reflect no selection. DS (9/7/01 3:03:59 PM) Seems to just be a general problem with the selection changed mechanism of the StyledText widget. StyledText.clearSelection(boolean) is being called with sendEvent == false? This results in no selection changed being fired to the listeners therefore there is no way for the actions to update. NOTES: DW (9/17/01 11:37:32 AM) File PR with workbench. DS (9/25/01 1:31:45 PM) Email from LK I think it is by design that we only send the selection event when the user performs the selection. I think this was based on how other SWT widgets handle selection (i.e., send it only when performed by the user vs. programmatically or as a side effect). Anyway, the behavior has been like that all along. Open a PR if you want different behavior. If we are inconsistent with the other widgets, that would be an argument for your case, but if not, we probably will not change the behavior. KR (9/26/01 11:18:42 AM) This is a bug in StyledText. A selection event should be sent anytime a user action (mouse click, key stroke) has changed the selection. Presently this does not work for any key strokes that modify the text content. It does not work in the backspace/delete scenario despite the fact that clearSelection(true) is called in invokeAction. The reason is that the selection is already changed/cleared in handleTextChanged which does not send a selection event (even though it should). DS (9/26/01 12:01:58 PM) Logged 1GKO6NY: SWT:ALL - Backspacing in StyledText does not fire selection changed DS (10/4/01 10:57:55 AM) From KR I released fixes for this PR and for "1GIK7D4: SWT:ALL - StyledText - Selection cleared when deletion disallowed by verify listeners". They didn't make it into build 203 but will be in the next.

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb212f'}

DS (9/10/01 10:55:55 AM) The consistency is between the Snippet Editor and the Java Editor. In the Java Editor the actions are in the popup menu: Inspect, Display ... why no Run the actions are not in the edit menu of the workbench In the SnippetEditor the actions are in the popup menu: Display, Inspect the actions are in the edit menu of the workbench NOTES: DW (9/17/01 11:41:14 AM) Actions should be ordered the same. We have no "run" in the debug context, as it is really just a snippet usage - i.e. run some code. Generally, when debugging, it is less useful to just "run" some code - usually you need to inspect a result. However, to be consistent, we should have the same options in both contexts, and we should add to the edit menu consistently.

minor

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2130'}

JGS (9/10/01 3:51:23 PM) Given the current way that view action delegates work, there is no way to reliably set the initial checked state of a contributred action. The workaround is to listen to the first SelectionChanged and set the checked state then. But really, this action has no business listening to SelectionChanged. Reconsider this problem when 2.0 is available, and see if there is any new infrastructure for contributed actions we could leverage. DW (9/21/2001 10:56:38 AM) This is the same/similar problem we have with the static/final filters. We do not persist the state of the static/final filters - we just set the initial state. Since the action delegate is not instantiated until the action (presentation) is toggled, there is no way to initialize properly. DW (9/24/2001 9:58:38 AM) Filed feature request with the workbench. 1GKKBCD: ITPUI:ALL - Feature: Dynamic/Persisted "state" for viewActions

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2131'}

EG (9/10/2001 10:18:52 AM) we now have JUnit launcher, that shows the results of a test run inside the current perspective. When the perspective switching preference is on and a test is run using the JUnit launcher then user a switched to the debug perspective, but the JUnit results view shows up in the Java perspective. Options: - the junit launcher can override the preference - the junit launcher can contribute its own switching preference - a launcher can query the preference setting and create the view in the target perspective if the switching preference is on (this requires workbench support). NOTES: DW (9/17/01 11:43:49 AM) Perhaps switching to the debug perspective should be part of a "launch cofiguration". EG (9/24/2001 8:41:20 AM) this would make sense but there should also be a sensible default.

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2132'}

PRODUCT VERSION: 135 Sun JDK1.2.2 JGS (9/10/01 5:22:48 PM) In InnerAnon scenario class, set breakpoints in nextElement() methods in both inner classes. Debug, and notice that the debugger is in a weird state where neither of the breakpoints was hit (as evidenced by console output), but DebugView still shows all threads as running with "unknown" labels. One of the breakpoints is shown as installed, the other isn't. If if you disable the line 38 breakpoint, you successfully stop at the line 25 breakpoint, but the program never terminates. If you try it the other way around, back to weird state. DS (9/13/01 3:56:37 PM) Cannot reproduce on 135 (or 135++). JGS (9/13/01 4:26:41 PM) Can still reproduce on 135 under Sun JDK1.2.2. Cannot reproduce under IBM VM or Sun JDK1.3. Under 1.2.2, following walkback appears: com.sun.jdi.VMDisconnectedException: Got IOException from Virtual Machine at org.eclipse.jdi.internal.connect.PacketReceiveManager.getReply(PacketReceiveManager.java:116) at org.eclipse.jdi.internal.connect.PacketReceiveManager.getReply(PacketReceiveManager.java:130) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:155) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:173) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:188) at org.eclipse.jdi.internal.ThreadReferenceImpl.resume(ThreadReferenceImpl.java:292) at org.eclipse.jdi.internal.event.EventSetImpl.resumeThreads(EventSetImpl.java:104) at org.eclipse.jdi.internal.event.EventSetImpl.resume(EventSetImpl.java:86) at org.eclipse.jdt.internal.debug.core.EventDispatcher.dispatch(EventDispatcher.java:104) at org.eclipse.jdt.internal.debug.core.EventDispatcher$1.run(EventDispatcher.java:136) at org.eclipse.core.internal.resources.Workspace.run(Workspace.java:1197) at org.eclipse.jdt.internal.debug.core.EventDispatcher.run(EventDispatcher.java:153) at java.lang.Thread.run(Thread.java:498) DS (9/13/01 4:54:15 PM) Not reliably reproduceable...takes me 1 in 10. Timing problem. Are we trying to talk to a VM too early? Appears to be a JDI problem. Should figure out what event is causing the problem. JGS (9/13/01 5:25:01 PM) Problem is a SocketException trying to read reply from telling thread to resume (JdwpCommandPacket is TR_RESUME). Added tracing to PacketManager to dump out each event. The first block is for a succesful run: Event=VMStartEvent: 0 Event=ThreadStartEvent: 5 Event=ClassPrepareEvent: 7 Event=ClassPrepareEvent: 11 Event=ClassPrepareEvent: 10 Event=ClassPrepareEvent: 11 Event=ClassPrepareEvent: 10 Event=BreakpointEvent: 13 Event=BreakpointEvent: 13 Event=BreakpointEvent: 13 Event=ClassPrepareEvent: 11 Event=ClassPrepareEvent: 10 Event=BreakpointEvent: 14 Event=BreakpointEvent: 14 Event=BreakpointEvent: 14 Event=ThreadDeathEvent: 6 Event=ThreadStartEvent: 5 Event=VMDisconnectEvent: 0 Next block is what we get when the problem happens: Event=VMStartEvent: 0 Event=ThreadStartEvent: 5 Event=ClassPrepareEvent: 7 Event=ClassPrepareEvent: 11 Event=ClassPrepareEvent: 10 Event=ClassPrepareEvent: 11 Event=ClassPrepareEvent: 10 com.sun.jdi.VMDisconnectedException: Got IOException from Virtual Machine at org.eclipse.jdi.internal.connect.PacketReceiveManager.getReply(PacketReceiveManager.java:116) at org.eclipse.jdi.internal.connect.PacketReceiveManager.getReply(PacketReceiveManager.java:130) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:155) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:173) at org.eclipse.jdi.internal.MirrorImpl.requestVM(MirrorImpl.java:188) at org.eclipse.jdi.internal.ThreadReferenceImpl.resume(ThreadReferenceImpl.java:292) at org.eclipse.jdi.internal.event.EventSetImpl.resumeThreads(EventSetImpl.java:104) at org.eclipse.jdi.internal.event.EventSetImpl.resume(EventSetImpl.java:86) at org.eclipse.jdt.internal.debug.core.EventDispatcher.dispatch(EventDispatcher.java:105) at org.eclipse.jdt.internal.debug.core.EventDispatcher$1.run(EventDispatcher.java:137) at org.eclipse.core.internal.resources.Workspace.run(Workspace.java:1197) at org.eclipse.jdt.internal.debug.core.EventDispatcher.run(EventDispatcher.java:154) at java.lang.Thread.run(Thread.java:498) So it appears the resume after a ClassPrepareEvent is causing the problem. JGS (9/13/01 5:43:56 PM) Source for InnerAnon follows: import java.util.*; public class InnerAnon { private Object[] stuff; private static String description = "foo"; private int answer = 42; public class Inner implements Enumeration { private int count; private Object[] contents; public Inner(Object[] cont) { contents = cont; count = 0; } public boolean hasMoreElements() { if (count < contents.length) { return true; } return false; } public Object nextElement() { return contents[count++]; // Set one breakpoint here } } public static Enumeration Anon(final Object[] cont) { return new Enumeration() { final int[] c = {0}; public boolean hasMoreElements() { return c[0] < cont.length; } public Object nextElement() { return cont[c[0]++]; // Set the second breakpoint here } }; } public InnerAnon(){ stuff = new Object[] {new String("one"), new String("two"), new String("three")}; } public static void main(String[] args) { InnerAnon ia = new InnerAnon(); Enumeration enum = ia.new Inner(ia.stuff); ia.printEnum(enum); enum = Anon(ia.stuff); ia.printEnum(enum); } public static void printEnum(Enumeration e) { final char lastLetter = 'z'; while (e.hasMoreElements()) { System.out.println((String)e.nextElement()); } } }

normal

RESOLVED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb2133'}

DS (9/12/01 3:27:20 PM) Launching with the following: C:\workspaces\eclipse\Testing>c:\jdk1.2.2\bin\java -cp "." -Xdebug -Xnoagent -Djava.compiler=NONE -Xrunjdwp:transport=dt_socket,server=y,suspend=y,address=8001 InfiniteLoop Attempting to disconnect the debug target results in FATAL ERROR in native method: Illegal ref passed to DeleteWeakGlobalRef abnormal program termination NOTES: DS (9/12/01 3:33:36 PM) Disconnect works on J9 and IBM vms...just not on the Sun VM 1.2.2 (1.3 works) java version "1.2.2" Classic VM (build JDK-1.2.2_005, native threads, symcjit)

critical

RESOLVED

WONTFIX

{'$oid': '52e9b49154dc1c25ebdb2134'}

The tool icon for "Show type names" in the variables view is currently a big X with some dots on a page. The big X and the page are ok, but the dots don't help - they mislead a bit - it makes me think that this tool might be for "Show qualified names" instead. (I always have to hover <g>). The X.x for "Show qualified names" is a bit better, although x.X would be more useful, given that package names are always in lowercase and type names are uppercase. I am thinking that these icons could be improved significantly by incorporating the standard 'type' icons - half green circle with C for class and half purple circle with I for interface - orange cube with a + through it for package Maybe if the X and the x.X were kept, and the "type" and "package" symbols were merged in, then these tool icons would be easier to differentiate. NOTES:

normal

VERIFIED

FIXED

{'$oid': '52e9b49154dc1c25ebdb2135'}

DS (9/13/01 10:55:18 AM) Create 2 projects, Testing and Testing2 In each project create a class called "Main" Create a main method in each class: public static void main(String[] args) { System.out.println(PROJECT_NAME); } Debug the Main in Testing Debug the Main in Testing2 The debug history menu correctly has two entries BUT both of the history items launch the Main in Testing. NOTES: MA (14.09.2001 11:30:54) me too! Maybe we can show full name in a tooltip, or show it as description in the status line?

minor

RESOLVED

DUPLICATE

{'$oid': '52e9b49154dc1c25ebdb2136'}

JGS (9/24/01 4:14:59 PM) Now that we have a PropertySheet for breakpoints, we should consider removing the context menu actions (Hit Count, Enable/Disable, etc.) from the BreakpointsView context menu. The main issue is multi-select behavior. Also see 1GKKEI5: ITPJUI:ALL - Small Feature: separate disable/enable breakpoints.

normal

RESOLVED

WONTFIX

{'$oid': '52e9b49254dc1c25ebdb2137'}

jkca (9/14/2001 2:39:31 PM) Currently, our code to open an editor when a Thread suspends does not take multipage editors into consideration. From a customer. What they want to do is to open their editor to the source page when a breakpoint is hit. They want to do this without activating the other pages. Our infrastructure does not support this. Our code that opens an editor and sets a marker to indicate the current line number does the following calls, taken from DebugView.openEditorAndSetMarker(...): <snip> if (editor == null) { if (fEditor != null) { if (!fEditor.isDirty()) { page.closeEditor(fEditor, false); } } try { editor= page.openEditor(input, editorId); fEditor = editor; page.activate(this); } catch (PartInitException e) { DebugUIUtils.logError(e); } } else { page.bringToTop(editor); } if (editor != null && (lineNumber >= 0 || charStart >= 0)) { //have an editor and either a lineNumber or a starting character IMarker marker= getInstructionPointer(lineNumber, charStart, charEnd, isTopStackFrame); editor.gotoMarker(marker); } </snip> Notice that we first call page.openEditor(input, editorId); and then later call editor.gotoMarker(marker);. We could just call page.openEditor(marker);, as is done in the TaskList. This change would put the responsibility of this support on the shoulders of the UI team. To solve the problem entirely, I believe the IWorkbenchPage API will have to be altered to take an argument to indicate which page the editor should open on. Also, we will have to ensure that we do not create/activate all pages in a multipage editor before they are required. On the other hand, the IWorkbenchPage API does have the facility to open an editor on a given marker. If the implementation of this was coded correctly, it would be enough to allow WSA to decide which page they would want to show when their multipage editor is first created. Similarly, we must not activate the other pages until they are needed. I have been unable to determine if the creation of multipage editors requires the creation & activation of all of the pages or just the current page. DS (9/14/01 5:05:59 PM) "We could just call page.openEditor(marker);" this can only be done if the marker is associated with a resource that is an IFile (@see WorkbenchPage#openEditor(IMarker, boolean). This does not handle the class file story (in jars).

normal

RESOLVED

DUPLICATE

{'$oid': '52e9b49254dc1c25ebdb2138'}

TM (9/17/01 12:39:29 PM) Eclipse JDI 116 Sometimes, while executing a test suite, I get this error: java.util.ConcurrentModificationException Stack trace: java/lang/Throwable.<init>()V java/util/Hashtable$HashIterator.next()Ljava/lang/Object; org/eclipse/jdi/internal/ValueCache.valuesWithType(Ljava/lang/Class;)Ljava/util/Collection; org/eclipse/jdi/internal/VirtualMachineImpl.resetEventFlagsOfAllThreads()V org/eclipse/jdi/internal/VirtualMachineImpl.resume()V com/oti/jdi/bench/framework/AbstractJDITest.runUntilReady()V com/oti/jdi/bench/framework/AbstractJDITest.setUp()V junit/framework/TestCase.runBare()V junit/framework/TestResult$1.protect()V junit/framework/TestResult.runProtected(Ljunit/framework/Test;Ljunit/framework/Protectable;)V junit/framework/TestResult.run(Ljunit/framework/TestCase;)V junit/framework/TestCase.run(Ljunit/framework/TestResult;)V junit/framework/TestSuite.run(Ljunit/framework/TestResult;)V junit/textui/TestRunner.doRun(Ljunit/framework/Test;Z)V junit/textui/TestRunner.run(Ljunit/framework/Test;)V junit/textui/TestRunner.run(Ljava/lang/Class;)V com/oti/jdi/bench/framework/JDITestRunner.run(Ljava/lang/Class;[Ljava/lang/String;)V com/oti/jdi/bench/tests/StepBench.main([Ljava/lang/String;)V TM (9/17/01 12:40:05 PM) The method in questions iterates over a vector without synchronizing on it.

normal

VERIFIED

FIXED

{'$oid': '52e9b49254dc1c25ebdb2139'}

jkca (9/14/2001 3:43:47 PM) Eclipse R0.9 Maintenance Stream There is a isomorphic relationship between debug models and model presentations. The following is a list of limitations that are created by this relationship: 1. Currently, the presentation is responsible for prompting the user for source. ISVs may want to use a different prompting method. Examples: JSP or XML debugging 2. The presentation may support transmogrifying the representation of information from one language to another. Examples: end to end debugging (viewing Javascript debug information as Java variables & stacks or as Javascript variables and stacks).

normal

VERIFIED

FIXED

{'$oid': '52e9b49254dc1c25ebdb213a'}

DW (9/24/2001 10:45:16 AM) When restoring breakpoints, we actively instantiate breakpoint classes. This will load the associated plug-ins. Not sure if this a break in the rules of engagement. EG (9/24/2001 8:20:30 AM) the editor has the same problem when it comes to Markers. We haven't found an easy solution.

normal

VERIFIED

FIXED

{'$oid': '52e9b49254dc1c25ebdb213b'}

In the Breakpoints view, there is a menu item: "Disable/Enable". Quite a number of times, there are a mix of enabled and disabled breakpoints and I want them all disabled but the only way to do that is sort out myself and disable them separately. Could you make it two menu items: Disable and Enable ? NOTES: EG (9/24/2001 8:10:45 AM) the toggle behaviour in the case of a multiple selection is not helpful. Suggest to add both Enable and Disable when the selection is a mix of enabled and disabled breakpoints. JGS (10/1/01 4:45:37 PM) Sent email to RG regarding multi-selection behavior in Breakpoints property sheet.

normal

VERIFIED

FIXED

{'$oid': '52e9b49254dc1c25ebdb213c'}

DW (10/10/2001 10:37:33 AM) Known problem for demo: Cannot perform an evaluation in a binary that is in a system package. This causes a "ClassNotFoundException". For example, in the debugger, stopped in "java.util.Vector.removeElement(Object)" and evaluating "size" causes something like: .java.lang.ClassNotFoundException: java/util/CodeSnippet_15 at java.lang.Class.forName1(Native Method) at java.lang.Class.forName(Class.java:134) at java.util.Vector.removeElement(Vector.java:601) at junit.samples.VectorTest.testRemoveElement(VectorTest.java:64) at java.lang.reflect.Method.invoke(Native Method) at junit.framework.TestCase.runTest(TestCase.java:166) at junit.framework.TestCase.runBare(TestCase.java:140) at junit.framework.TestResult$1.protect(TestResult.java:106) at junit.framework.TestResult.runProtected(TestResult.java:124) at junit.framework.TestResult.run(TestResult.java:109) at junit.framework.TestCase.run(TestCase.java:131) at junit.framework.TestSuite.runTest(TestSuite.java:173) at junit.framework.TestSuite.run(TestSuite.java:168) at junit.textui.TestRunner.doRun(TestRunner.java:74) at junit.textui.TestRunner.run(TestRunner.java:200) at junit.samples.VectorTest.main(VectorTest.java:19)

normal

RESOLVED

DUPLICATE