Task tags in comments are not detected in some cases

[travkin79]

When working with C/C++ code using CDT LSP (which uses the C/C++ grammar and language configuration from TM4E), I found some cases where task tags are not detected in C++ comments where I would expected them to be detected.

Steps to reproduce

1. Install CDT LSP 3.4 and TM4E 0.17.1
2. Take the following C++ code example and open it the LSP-based C++ editor
3. Check the Tasks view and the markers on the left side of the editor

// TODO task 1
class SomeClass
{
    // TODO task 2
public:

  // TODO task 3
  void f1() // TODO task 4
  {
    // TODO task 5
    
    if (true) // TODO task 6
    {
        // TODO task 7
        bool value = false; // TODO task 8
        // TODO task 9
    }
    
    // TODO task 10
  }
  
  // TODO task 11
};

// TODO task 12

Expected result: All tasks (1 - 12) should be listed in the Tasks view and should be marked on the left side of the editor.

Actual result: Only tasks 1, 2, 3, 11, 12 are listed in the tasks view and marked in the editor, see Screenshot:

Environment

• TM4E 0.17.1
• CDT LSP 3.4
• Eclipse 2025-12
• RHEL 9.2

Remarks

It seems, the issue has something to do with https://github.com/eclipse-tm4e/tm4e/blob/main/org.eclipse.tm4e.ui/src/main/java/org/eclipse/tm4e/ui/internal/utils/MarkerUtils.java. While debugging, I found out that the comment tokens are found, but the following condition in updateTextMarkers method is not met since the token type contains the definition string. Maybe the second part of this condition is wrong or there is something wrong with the C/C++ grammar.

if (!token.type.contains("comment") || token.type.contains("definition"))

[sebthom]

I guess the C grammar wrongly classifies the actual comment content as definition which usually is used to indicate the comment start/end tokens e.g. //, /* etc.

For example in JS the actual line comment is classified as js.line.comment.double-slash and the // as punctuation.definition.comment.js

At the end there is no definitive rule to for the IDs and the MarkerUtils use heuristic but since it is common practice to use definition for the start/end tokens and not for the comments I am not really motivated to touch that code and would suggest you raise an issue against the C grammar repo to sort this out.

[travkin79]

Hi @sebthom,
Thank you for the explanation. Could you please point me to the C/C++ grammar repo? Then I'll create an issue there.

Is it the following git repo?
https://github.com/jeff-hykin/better-cpp-syntax

[travkin79]

Hi @sebthom,
I've created an issue in the C/C++ grammar git repo. I'm not sure if I correctly described the issue. Could you please take a look?
jeff-hykin/better-cpp-syntax#693
Thank you.

[sebthom]

Looks good, let's see what they say

[sebthom]

We are using the C++ syntax definition sourced from VS Code (https://github.com/microsoft/vscode/tree/main/extensions/cpp), which is an older but MIT-licensed version of https://github.com/jeff-hykin/better-cpp-syntax/.

Any fixes added upstream will therefore only appear in the majorly refactored and GPL-licensed version and will not be usable for us.

[travkin79]

Hi @sebthom,

We are using the C++ syntax definition sourced from VS Code (https://github.com/microsoft/vscode/tree/main/extensions/cpp), which is an older but MIT-licensed version of https://github.com/jeff-hykin/better-cpp-syntax/.

Any fixes added upstream will therefore only appear in the majorly refactored and GPL-licensed version and will not be usable for us.

So, how do improvements / bug fixes in the C/C++ grammar find their way to VSCode and TM4E usually?

[sebthom]

So, how do improvements / bug fixes in the C/C++ grammar find their way to VSCode and TM4E usually?

We strictly sync the vscode syntaxes. we don't maintain any alterations. So issues would need to be reported against vscode repository. But I doubt this will be resolved.

Alternatively, the CDT team could register an updated TextMate syntax with higher priority in their plugin.xml