Uncaught exception: 'java.project.listSourcePaths' already exists

[davidpcaldwell]

When developing my own extension and running it under the Extension Development Host in the debugger (I have it set to pause on uncaught exceptions), on opening a Java project (with this extension installed), I am getting the following stack trace:

Exception has occurred: Error: command 'java.project.listSourcePaths' already exists
t u.registerCommand ([/Users/inonit/Applications/Visual]() Studio Code.app[/Contents/Resources/app/out/vs/workbench/services/extensions/node/extensionHostProcess.js:93:28683]())
	at Object.registerCommand ([/Users/inonit/Applications/Visual]() Studio Code.app[/Contents/Resources/app/out/vs/workbench/services/extensions/node/extensionHostProcess.js:107:30423]())
	at X.register ([/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:945419]())
	at [/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:966900]()
	at new Promise (<anonymous>)
	at w.handleRegistrationRequest ([/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:966543]())
	at [/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:957330]()
	at [/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:1106102]()
	at [/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:1107161]()
	at Immediate._onImmediate ([/Users/inonit/.vscode/extensions/redhat.java-1.3.0/dist/extension.js:2:1109867]())
	at processImmediate (internal[/timers.js:461:21]())
	at process.callbackTrampoline (internal[/async_hooks.js:131:14]())```

##### Environment
- Operating System: macOS (10.15.7 and 12.2.1)
- JDK version: Oracle JDK 1.8.0_212 and AdoptOpenJDK 11.0.8
- Visual Studio Code version: 1.64.2
- Java extension version: 1.3.0

##### Steps To Reproduce
1. Open Extension Development Host with my (unpublished) extension and with a window containing Java projects
2. Wait. I don't know the API well enough to know whether this is on a timer (the stack trace may suggest this) or whether it's more complicated than that (`processImmediate` is maybe just something to do with the event loop)?

##### Current Result
See stack trace of uncaught exception above.

##### Expected Result
No uncaught exception.

##### Additional Informations
![Screen Shot 2022-02-23 at 2 26 38 PM](https://user-images.githubusercontent.com/1809061/155395932-3a03d2f6-e15e-432d-8866-12721772c8b3.png)

[rgrunber]

I see it as well

Exception has occurred: Error: command 'java.project.listSourcePaths' already exists
t u.registerCommand ([/usr/share/code/resources/app/out/vs/workbench/services/extensions/node/extensionHostProcess.js:93:28683]())
	at Object.registerCommand ([/usr/share/code/resources/app/out/vs/workbench/services/extensions/node/extensionHostProcess.js:107:30423]())
	at ExecuteCommandFeature.register ([/home/rgrunber/git/vscode-java/dist/extension.js:79361:52]())
	at [/home/rgrunber/git/vscode-java/dist/extension.js:80279:29]()
	at new Promise (<anonymous>)
	at LanguageClient.handleRegistrationRequest ([/home/rgrunber/git/vscode-java/dist/extension.js:80265:16]())
	at [/home/rgrunber/git/vscode-java/dist/extension.js:79913:108]()
	at handleRequest ([/home/rgrunber/git/vscode-java/dist/extension.js:89822:41]())
	at processMessageQueue ([/home/rgrunber/git/vscode-java/dist/extension.js:89686:17]())
	at Immediate._onImmediate ([/home/rgrunber/git/vscode-java/dist/extension.js:89676:13]())
	at processImmediate (internal[/timers.js:461:21]())
	at process.callbackTrampoline (internal[/async_hooks.js:131:14]())

It seemed the stacktrace was associated with the language server registering commands to the client side, so I recalled we have https://github.com/eclipse/eclipse.jdt.ls/blob/master/org.eclipse.jdt.ls.core/plugin.xml#L54-L106 in JDT-LS. Any command there that's also registered on the client side (eg. https://github.com/redhat-developer/vscode-java/blob/master/src/buildpath.ts#L48) would run into this.

@fbricon @snjeza , is there a reason we register in both places ?

Looks to me like we have a command on the client side that is just meant to call the same command on the server side. Maybe just using a slightly different name would avoid the clash.

[rgrunber]

I don't think the exception results in any unwanted behaviour, though I have a simple patch to fix it. I think the issue happens as follows :

We register a command on the client-side (eg. https://github.com/redhat-developer/vscode-java/blob/master/src/buildpath.ts#L48-L49) that simply calls java.execute.workspaceCommand on the server side with the same named command id. That command id is also registered by the server to the client. See https://github.com/eclipse/eclipse.jdt.ls/blob/master/org.eclipse.jdt.ls.core/src/org/eclipse/jdt/ls/core/internal/handlers/JDTLanguageServer.java#L350-L351 & https://github.com/eclipse/eclipse.jdt.ls/blob/master/org.eclipse.jdt.ls.core/plugin.xml#L74 .

The client side is registered first and the server side fails to register its command (already exists). However, I don't think this is an issue because we call the server side commands through `java.execute.workspaceCommand , and not directly.

Update : Yup, https://github.com/microsoft/vscode-languageserver-node/blob/main/client/src/common/client.ts#L2887-L2891 handles the registration of the commands passed to workspace/executeCommand coming from the server. It registers a command with the same ID. So one could do vscode.commands.executeCommand(ID) or vscode.commands.executeCommand("java.workspace.executeCommand", ID) and both should work.