Blueprint "event begin play" called before C++ parent's "event begin play"?

Development Programming & Scripting C++
, , , , ,
23

I believe I had the same issue and made an example for it:
test_pc.h:

#pragma once

#include "CoreMinimal.h"
#include "GameFramework/PlayerController.h"
#include "test_pc.generated.h"


UCLASS()
class MY_GAME_API Atest_pc : public APlayerController {
	GENERATED_BODY()

protected:
    void my_log_to_screen(const char* msg) {
        if (GEngine) {
            GEngine->AddOnScreenDebugMessage(-1, 20.f, FColor::White, msg);
        }
    }
	virtual void BeginPlay() override {
		my_log_to_screen("BeginPlay call started");
		Super::BeginPlay();
		my_log_to_screen("Super of BeginPlay finished");
		if (IsLocalPlayerController()) {
			OnPossessedPawnChanged.AddDynamic(this, &Atest_pc::process_pawn_change);
		    my_log_to_screen("Bind finished");
		}
    }

    UFUNCTION()
	void process_pawn_change(APawn* old_pawn, APawn* new_pawn) {
		my_log_to_screen("Process pawn change called");
	}
};

bp_test_pc:

image
image1072×317 84.6 KB

bp_test_gm:

image
image872×324 61.1 KB

output:
image
From the output it seems like execution order is reversed.

Adding a delay:

image
image1067×375 95.1 KB

Or in the C++ class moving the Super::BeginPlay() at the end of the BeginPlay() function solves the order issue:
image

But as a software engineer this looks terribly wrong. Using another function other than BeginPlay() feels like a workaround for a bug in BeginPlay (same for moving it the Super::BeginPlay() to the end), as BeginPlay is widely used in tutorials (including Epic’s) and also in generated c++ classes.
This looks like a bug to me.