Using dot '.' as member access operator fails to trigger auto completion/intellisense when used inside macro.

[mehlian]

Environment

• OS and Version: Windows 10
• VS Code Version: 1.108.2
• C/C++ Extension Version: 1.32.2
• If using SSH remote, specify OS of remote machine: none

Bug Summary and Steps to Reproduce

Bug Summary:
There is a feature that allows to type '.' as access member operator on any type to show completion window with all members and then it resolves to '.' or '->' automatically.
This feature fails when used inside macro call.

Steps to reproduce:

1. Paste code (file.hpp):

#define myMacro(value) 

struct MyStruct
{
    int value;
};

struct Container
{
    int value2;
    MyStruct *myStruct;
};

void func(Container *container)
{
    myMacro(container.                         // <- problem here
}

2. Go to 'func' function and type in 'myMacro(container'
3. Now when we type access member operator '.' normally popup with suggestions should appear - but it doesn't.
4. Instead of '.' type '->' - suggestions appear.

Expected behavior:
After typing '.' in macro call we should see suggestions with struct members just like after typing '->'.
'Trigger Suggest' command (CTRL+SPACE) doesn't work on '.' but works after '->' when used in macro call.

Configuration and Logs

ask for more details

Other Extensions

No response

Additional context

No response

[sean-mcmanus]

@mehlian This is caused by our completion implementation hitting an issue due to the macro not being completed by a ). During normal typing of myMacro( the ) should by default be auto-completed giving the expected results:

Is that sufficient or did you really want the incomplete macro case to work? FYI, that case frequently works via Copilot's autocomplete:

[mehlian]

I would like the incomplete macro case to work. This annoys me for over 2 years now.

[sean-mcmanus]

@mehlian I filed a bug on our shared VS component at https://developercommunity.visualstudio.com/t/C-IntelliSense-completion-fails-after-/11126235 , but it's not publicly visible yet (not sure when it will be). We might also be able to fix it ourselves, although we're currently trying to focus on fixing the known regressions blocking our 1.33.x release.

UPDATE: It seems like it should be visible now (unless there's some lag I'm not aware of).