David Heffernan

Then please provide a complete but minimal reproduction.

This is just a debugger issue, you shouldn't change your code because of this

The best piece of advice you can take from here is that you need to learn how to ask questions effectively. Because if you ask like this then nobody can help you. And it's just a waste of your time.

No, platform tests are important and valuable. But framework dependency tests have no power here because you can read a single uses clause and know it's fine in 15s.

No library developer would test things like this. I mean why stop at FMX/VCL? What about database frameworks. Does it have dependencies on any of them? Does it work in a Windows service? Should there be a test for that? It's simple to see by inspecting the uses clause, so the developer just does not need to do any of that.

How could VCL/FMX be relevant to code at the RTL level?

Where does the exception get raised?

OK so your view is not to raise exceptions at all, in which case Delphi is the wrong language for you because you can't avoid dealing with them. But this question is based on the premise that exceptions are used so responding with this viewpoint seems a little off topic.

Why bother with exceptions then at all? This is just back to error handling without exceptions where every single function call returns a status code or flag, like calling a Win32 function.

Handle them as little as possible is the rule you need. Only mutate the exception if you need to.

You are most welcome! 😉

That's not what this is about, generics being a bottleneck. The potential bottleneck is the RTL dictionary class. Which happens to be a generic type.

Nice thought experiment

And yet you did this with a other library? I'm curious. Why was it simple to integrate rapid? Does it have the same interface as rtl but is just faster?

Seeing this removed all my interest in helping. You don't have a clue what you are doing so blame Microsoft rather than yourself.