Integrate rust-analyzer and improve LSP Integration (required function, guards for empty replies, icons, deadlock fix)

[matthiasblaesing]

General improvements (lsp.client)

To be able to sanely use rust-analyzer (the rust LSP implementation) the NetBeans lsp client implementation needs some adjustments:

• lsp4j expects client implementation to provide at least a dummy implementation of refreshDiagnostics, failing to correctly work if that is not the case
• it was observed that the rust lsp in many cases returns null as a result of LSP operations. The existing null guards were thus extended
• icons in the lsp.client module were referenced but not actually present causing exceptions at runtime when they were to be used.
• it was observed, that requests to apply document formatting might block if the language servers issues a publishDiagnostics call as part of the formatting.
• The CSL task was executed independently of the presence of its prerequisites, overwriting the results of other tasks if necessary.

Rust specific

The cargo TOML support is updated to the current version of TOML to be able to parse project files using extended syntax.

While NetBeans can provide parts of the IDE services based on custom implementations, the rust-analyzer LSP server provides extended support of the box. Compared with the built-in solution rust-analyzer has the drawback, that it is huge and platform dependent, so it can't be bundled with NetBeans.

To provide both minimal rust support in all cases and improved editing support by using rust-analyzer the solution provided here will switch from internal support to rust-analyzer only if rust-analyzer is installed and configured.

When a user now opens a rust file and rust-analyzer is not configured a warning is attached to the file (as is done for cpplite):

But structure is still present:

The icons used in the rust module were aligned with the other NetBeans icons from lsp.client and PHP support.

The warning allows to directly jump into options to configure rust-analyzer:

With that done extended structure support is activated:

Code completion is provided:

and Find usages works:

[matthiasblaesing]

@vieiro could you please have a general look?

@lahodaj could you please have look at the general LSP and CSL changes and see if they make sense to you?

[vieiro]

Hi @matthiasblaesing ,

Thanks for this effort!

I wasn't able to get LSP server running properly with:

$ which rust-analyzer 
~/.cargo/bin/rust-analyzer
$ rust-analyzer --version
rust-analyzer 1.90.0 (1159e78 2025-09-14)

The NetBeans UI froze, with the EDT trying to update nodes [1].

I haven't found time to look at it in detail, but from [2] I wonder if we should be using SwingUtilities.invokeLater in the setKeys near org.netbeans.modules.lsp.client.bindings.NavigatorPanelImpl.run(NavigatorPanelImpl.java:77)

I think it's premature to add this to the IDE right now. Maybe we want to iterate a little bit more.

Kind regards,
Antonio

[1]

   java.lang.Thread.State: RUNNABLE
	at org.openide.util.Mutex.isReadAccess(Mutex.java:313)
	at org.openide.nodes.EntrySupportDefault.getArray(EntrySupportDefault.java:673)
	at org.openide.nodes.EntrySupportDefault$Info.nodes(EntrySupportDefault.java:781)
	at org.openide.nodes.EntrySupportDefault.justComputeNodes(EntrySupportDefault.java:172)
	at org.openide.nodes.ChildrenArray.nodes(ChildrenArray.java:63)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:107)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:149)
	at org.openide.nodes.Children.getNodes(Children.java:444)
	at org.openide.nodes.Node.assignTo(Node.java:318)
	- locked <0x0000000421bd4f88> (a java.lang.Object)
	at org.openide.nodes.EntrySupportDefault.justComputeNodes(EntrySupportDefault.java:189)
	at org.openide.nodes.ChildrenArray.nodes(ChildrenArray.java:63)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:107)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:149)
	at org.openide.nodes.Children.getNodes(Children.java:444)
	at org.openide.nodes.Node.assignTo(Node.java:318)
	- locked <0x0000000421bd4f88> (a java.lang.Object)
	at org.openide.nodes.EntrySupportDefault.justComputeNodes(EntrySupportDefault.java:189)
	at org.openide.nodes.ChildrenArray.nodes(ChildrenArray.java:63)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:107)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:149)
	at org.openide.nodes.Children.getNodes(Children.java:444)
	at org.openide.nodes.Node.assignTo(Node.java:318)
	- locked <0x0000000421bd4f88> (a java.lang.Object)
	at org.openide.nodes.EntrySupportDefault.justComputeNodes(EntrySupportDefault.java:189)
	at org.openide.nodes.ChildrenArray.nodes(ChildrenArray.java:63)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:107)
	at org.openide.nodes.EntrySupportDefault.getNodes(EntrySupportDefault.java:149)
	at org.openide.nodes.Children.getNodes(Children.java:444)
	at org.openide.nodes.Node.assignTo(Node.java:318)

[2]

	at org.openide.nodes.Children$Keys.applyKeys(Children.java:1556)
	at org.openide.nodes.Children$Keys.setKeys(Children.java:1500)
	at org.netbeans.modules.lsp.client.bindings.NavigatorPanelImpl.run(NavigatorPanelImpl.java:77)
	at org.netbeans.modules.lsp.client.LSPBindings.lambda$addBackgroundTask$19(LSPBindings.java:563)
	at org.netbeans.modules.lsp.client.LSPBindings$$Lambda/0x00000000689ee8f0.run(Unknown Source)

[vieiro]

The thread dump.txt I was able to extract when the IDE froze.

[matthiasblaesing]

I think it's premature to add this to the IDE right now. Maybe we want to iterate a little bit more.

I'll look into the points you brought up. BUT I have to ask: for what do we want to wait? Rust Support in NetBeans has not moved an inch and while this is rough, not doing it will not yield improvements.

[matthiasblaesing]

Where does that version of rust-analyzer come from? I looked through

https://github.com/rust-lang/rust-analyzer/releases

and

https://github.com/rust-lang/rust-analyzer/tags

there is neither a tag, nor a release entry in september of last year. I fail to find the reference 1159e78 in the rust-analyzer repository (it does not resolve to a commit). The latest version of rust-analyzer is 0.3.2811, so that also makes no sense.

[matthiasblaesing]

@vieiro could you please describe how you reached the deadlock and whether that is reproducible?

[vieiro]

I think it's premature to add this to the IDE right now. Maybe we want to iterate a little bit more.

I'll look into the points you brought up. BUT I have to ask: for what do we want to wait? Rust Support in NetBeans has not moved an inch and while this is rough, not doing it will not yield improvements.

Very true! If this is on master we'll have more chances to get contributions!

[vieiro]

@vieiro could you please describe how you reached the deadlock and whether that is reproducible?

I had a project opened (and a Rust file opened in the editor). Then the IDE got restarted (after changing the rust-analyzer path) and then it happened. I think there's a race condition between the LSP server being started and the opened files being sent to the LSP server, or something like that.

IIRC there were two rust-analyzer servers running at the same time. I'll try to get a reproducer, though.

[vieiro]

Hi @matthiasblaesing ,

I was trying today again with different projects and couldn't reproduce the IDE freeze. I've started the rust-analyzer with these options to have more details on what's going on. No need to include these in the commit, I'm posting them here just in case they're of interest in the future.

diff --git a/rust/rust.grammar/src/org/netbeans/modules/rust/grammar/lsp/RustLSP.java b/rust/rust.grammar/src/org/netbeans/modules/rust/grammar/lsp/RustLSP.java
index 0a8bdda40c..c1b2aee366 100644
--- a/rust/rust.grammar/src/org/netbeans/modules/rust/grammar/lsp/RustLSP.java
+++ b/rust/rust.grammar/src/org/netbeans/modules/rust/grammar/lsp/RustLSP.java
@@ -39,16 +39,23 @@ public class RustLSP implements LanguageServerProvider {
     public LanguageServerDescription startServer(Lookup lookup) {
         Path rustAnalyzerPath = RustAnalyzerOptions.getRustAnalyzerLocation(true, true);
         if(rustAnalyzerPath == null || ! Files.isExecutable(rustAnalyzerPath)) {
+            LOG.log(Level.INFO, String.format("rustAnalyzerPath is not available: %s", rustAnalyzerPath));
             return null;
         }
         try {
-            Process p = new ProcessBuilder(new String[]{rustAnalyzerPath.toAbsolutePath().toString()})
+            Process p = new ProcessBuilder(new String[]{
+                rustAnalyzerPath.toAbsolutePath().toString(),
+                    "--verbose",
+                    "--log-file",
+                    "/tmp/rust-analyzer.log",
+                    "--no-log-buffering"
+            })
                     .directory(rustAnalyzerPath.getParent().toFile())
                     .redirectError(ProcessBuilder.Redirect.INHERIT)
                     .start();
             return LanguageServerDescription.create(p.getInputStream(), p.getOutputStream(), p);
         } catch (IOException ex) {
-            LOG.log(Level.FINE, null, ex);
+            LOG.log(Level.INFO, String.format("Error launching rust-analyzer: %s", ex.getMessage()), ex);
             return null;
         }
     }

I found some other issues, though, but I think we can work out these in the future. I'll post them here for future reference:

     [exec] WARNING [org.netbeans.modules.versioning.util.Utils]: associateEncoding() no file object available for /tmp/vcs-1773087075761/vcs-1773087079972/lib.rs
     [exec] SEVERE [org.openide.util.Exceptions]
     [exec] org.eclipse.lsp4j.jsonrpc.ResponseErrorException: content modified
     [exec] 	at org.eclipse.lsp4j.jsonrpc.RemoteEndpoint.handleResponse(RemoteEndpoint.java:220)
     [exec] 	at org.eclipse.lsp4j.jsonrpc.RemoteEndpoint.consume(RemoteEndpoint.java:204)
     [exec] 	at org.eclipse.lsp4j.jsonrpc.json.StreamMessageProducer.handleMessage(StreamMessageProducer.java:185)
     [exec] 	at org.eclipse.lsp4j.jsonrpc.json.StreamMessageProducer.listen(StreamMessageProducer.java:97)
     [exec] 	at org.eclipse.lsp4j.jsonrpc.json.ConcurrentMessageProcessor.run(ConcurrentMessageProcessor.java:114)
     [exec] 	at java.base/java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:545)
     [exec] 	at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:328)
     [exec] 	at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1090)
     [exec] 	at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:614)
     [exec] 	at java.base/java.lang.Thread.run(Thread.java:1474)
     [exec] Caused: java.util.concurrent.ExecutionException
     [exec] 	at java.base/java.util.concurrent.CompletableFuture.wrapInExecutionException(CompletableFuture.java:345)
     [exec] 	at java.base/java.util.concurrent.CompletableFuture.reportGet(CompletableFuture.java:440)
     [exec] 	at java.base/java.util.concurrent.CompletableFuture.get(CompletableFuture.java:2094)
     [exec] [catch] at org.netbeans.modules.lsp.client.Utils.handleBindings(Utils.java:403)
     [exec] 	at org.netbeans.modules.lsp.client.Utils.handleBindings(Utils.java:352)
     [exec] 	at org.netbeans.modules.lsp.client.bindings.MarkOccurrences.computeHighlights(MarkOccurrences.java:136)
     [exec] 	at org.netbeans.modules.lsp.client.bindings.MarkOccurrences.run(MarkOccurrences.java:106)
     [exec] 	at org.netbeans.modules.lsp.client.LSPBindings.lambda$addBackgroundTask$17(LSPBindings.java:550)
     [exec] 	at org.openide.util.RequestProcessor$Task.run(RequestProcessor.java:1403)
     [exec] 	at org.netbeans.modules.openide.util.GlobalLookup.execute(GlobalLookup.java:45)
     [exec] 	at org.openide.util.lookup.Lookups.executeWith(Lookups.java:287)
     [exec] 	at org.openide.util.RequestProcessor$Processor.run(RequestProcessor.java:2012)

And

     [exec] SEVERE [org.openide.util.RequestProcessor]: Error in RequestProcessor org.netbeans.modules.lsp.client.LSPBindings$$Lambda/0x000000000d8c1d60
     [exec] java.lang.IllegalArgumentException: Invalid end offset: 3203, start is: 3210
     [exec] 	at org.netbeans.spi.editor.fold.FoldInfo.<init>(FoldInfo.java:103)
     [exec] 	at org.netbeans.spi.editor.fold.FoldInfo.range(FoldInfo.java:94)
     [exec] 	at org.netbeans.modules.lsp.client.bindings.FoldManagerImpl.computeInfos(FoldManagerImpl.java:170)
     [exec] 	at org.netbeans.modules.lsp.client.bindings.FoldManagerImpl.lambda$run$3(FoldManagerImpl.java:125)
     [exec] 	at org.netbeans.modules.lsp.client.Utils.handleBindings(Utils.java:405)
     [exec] 	at org.netbeans.modules.lsp.client.Utils.handleBindings(Utils.java:352)
     [exec] 	at org.netbeans.modules.lsp.client.bindings.FoldManagerImpl.run(FoldManagerImpl.java:121)
     [exec] 	at org.netbeans.modules.lsp.client.LSPBindings.lambda$addBackgroundTask$17(LSPBindings.java:550)
     [exec] 	at org.openide.util.RequestProcessor$Task.run(RequestProcessor.java:1403)
     [exec] 	at org.netbeans.modules.openide.util.GlobalLookup.execute(GlobalLookup.java:45)
     [exec] 	at org.openide.util.lookup.Lookups.executeWith(Lookups.java:287)
     [exec] 	at org.openide.util.RequestProcessor$Processor.run(RequestProcessor.java:2012)
     [exec] Caused: org.openide.util.RequestProcessor$FastItem

Thanks for the PR. I'm approving changes now!

[vieiro]

Good to go. Let's get this merged-in, and improve in the future.

[vieiro]

FWIW, having the rust-analyzer use a log file may be of interest, because it prints-out interesting stuff (maybe it hangs too?). This could be done using a System.property pointing to a log file (say netbeans.rust.analyzer.log_file or something). Who knows, maybe it's the rust-analyzer process that is hanging causing GUI freeze.

For instance, if you mouse+click on a Rust Result in the editor, then the corresponding Result.rs is opened (from the Rust core project). Then rust-analyzer generates some errors in the rust-analyzer.log file:

$ cat /tmp/rust-analyzer.log 
2026-03-09T22:19:58.76232392+01:00 ERROR failed fetching cargo workspace root e=cd "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core" && "/home/antonio/.cargo/bin/cargo" "locate-project" "--workspace" "--manifest-path" "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/Cargo.toml" failed, exit status: 101
stderr:
error: failed to parse manifest at `/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/Cargo.toml`

Caused by:
  the cargo feature `profile-rustflags` requires a nightly version of Cargo, but this is the `stable` channel
  See https://doc.rust-lang.org/book/appendix-07-nightly-rust.html for more information about Rust release channels.
  See https://doc.rust-lang.org/cargo/reference/unstable.html#profile-rustflags-option for more information about using this feature.
 cargo_toml=/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/Cargo.toml
2026-03-09T22:20:00.679405718+01:00 ERROR failed fetching cargo workspace root e=cd "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core" && "/home/antonio/.cargo/bin/cargo" "locate-project" "--workspace" "--manifest-path" "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/Cargo.toml" failed, exit status: 101
stderr:
error: failed to parse manifest at `/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/Cargo.toml`

Caused by:
  the cargo feature `profile-rustflags` requires a nightly version of Cargo, but this is the `stable` channel
  See https://doc.rust-lang.org/book/appendix-07-nightly-rust.html for more information about Rust release channels.
  See https://doc.rust-lang.org/cargo/reference/unstable.html#profile-rustflags-option for more information about using this feature.
 cargo_toml=/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/Cargo.toml
2026-03-09T22:20:01.200179926+01:00 ERROR failed fetching cargo workspace root e=cd "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core" && "/home/antonio/.cargo/bin/cargo" "locate-project" "--workspace" "--manifest-path" "/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/Cargo.toml" failed, exit status: 101
stderr:
error: failed to parse manifest at `/home/antonio/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/Cargo.toml`

The process keeps on running well but who knows, maybe it causes a GUI freeze in other circumstances...

[matthiasblaesing]

@vieiro thank you for your input. For the FoldInfo case it would be helpful to know whether this is reproducible and if so, how. I can imagine, that LSPs report bogus information and if so we should filter that, but an example would be helpful to examine this.

I intent to keep this PR open for a few days and merge early next week unless concerns are raised.

[lahodaj]

I looked at the LSP client changes, makes sense to me.

I wonder if there's a way to simplify/streamline the null checks (sadly, the return type/value is not an Optional), but that's to a great deal a separate concern.