Devajith

Windows is not happy with the new test. I'm not sure how windows handles errors/warnings. Here's how the warning message looks now: 

A stripped down version of the header that segfaults: ``` // test.h #include "LoKi/Particles.h" using EQUALTO = LoKi::EqualToValue; const auto TRTYPE = LoKi::Particles::TrackType{}; // Remove ANY of the lines below...

> Is there a corresponding roottest PR enabling/adding test for this? Added a cling test for this.

Not a proper solution. What we ideally want to do is check if a `VarDecl` with the same identifier is already added for the current `ControlScope`.

Minimal reproducer: ``` if(true) {auto h1 = new int(any_undeclared_variable_here);} ```

The above example throws an expected error in `cling`: ``` [cling]$ auto h1 = new int(any_undeclared_variable_here); input_line_5:2:20: error: use of undeclared identifier 'any_undeclared_variable_here' auto h1 = new int(any_undeclared_variable_here); ^ ```...

As @jalopezg-git mentioned, bool(foo) appears as a declaration (bool foo). The following simple C++ code also fails to compile with a re-declaration error and I think that should be the...

As a note, once we remove `auto auto` support in: https://github.com/root-project/root/pull/16410, this will no longer be needed and will need to be reverted.

It only happens for this specific case `for(i=0; i

I was looking into this and this seems similar to the issue: https://github.com/root-project/root/issues/10049 It's strange that this works: ``` root [0] std::unique_ptr foo{TFile::Open("foo.root", "RECREATE")}; root [1] foo.GetName() ROOT_prompt_1:1:4: error: no...