What i've understood is that they have native speed while exporting on ios (which is done with one command by the way).
I am not against visual interface and help from the framework. But to give an example, I had to abandon 2 projects because one of them required complex AI logic which was a way too heavy to do with events. It was an action script seek and kill with multiple floors with ladders and multiple enemies.
It took about 2 hours to build a logic, which was buggy by the way because C2's behaviors didn't fit the style, and most of the time was spent on clicking. You may say yeah you can create your personnal plugins and stuff. Well okay but basically it should not happen, C2 is too unspecific (or maybe just too specific to suit all of my needs)
Another bug was happening with physics behavior where you had to move a ball in a labyrinth. Sometimes you've had to rotate at 90° the labyrinth with the ball in it and strangely the ball was falling through occasionally.
Another one where i wanted to use pathfinding behavior with disabled diagonals pathfinding, well in the beginning this went as it should but the movement wasn't straight at the end. The enemy was turning a bit, which my client didn't want. And programming all this with events is a pain in the ass.
If the behaviors aren't working for you, you can customize the behavior with events. You can also use events rather than needing to go as far as making your own plugin. If you think making it with events instead of behaviors is difficult, writing it all from scratch in code is much harder. Also, two hours on ai really is not much at all. AI is notoriously difficult no matter what you code with. AAA games with millions of dollars put into them sometimes struggle with it too. It is extremely likely that C2 is capable of what you're trying to do, but you're trying to tackle something more difficult than you can manage with your current level of experience with C2. I did the same thing with the first version of my rpg, loot pursuit.
Recent project. Where the collision with a trigger should have been spawning an object and moving it along other ones to make the transition seamless. Well there were gaps, that meaned that construct doesn't have a good collision detection for this purpose.
I posted in your thread, that problem was incorrect math, it didn't have anything to do with collision detection.
I mean common you can do simple games. But it's a pain in the ass for big ones and complicated ( physically, logically..)
C2 is plenty capable of making large and complex games, I've been making two and it works fine for them. It does require knowing how to use the program well, though, but the same will be true of whatever framework you use. Large complex projects need to be designed correctly to remain manageable in C2, C++ and everything else.
You might like typing code more than making events, but I assure you, you will encounter similar problems writing code directly.