Yeah, I definitely get your point and I mostly agree. And I probably don't want to write a full RPerl application.
But if I could write my main program in Perl and have my limited CPU intensive part(s) of the code (e.g. deeply nested loops) in RPerl, maybe I could have the best of two worlds: high speed of compiled programs for the CPU intensive part, and the high expressive power of Perl for the rest of the program.
I once agree with you, but now I don't. you think RPerl is not real perl, just because rperl is not smart enough yet.
I don't talk with Will Braswell, But I believe that his target is Rperl will be a compiler for perl which compile all no magic perl code but remain magic perl for perl itself automatically!. That would save a lot of time to manual write C/XS code for performance.
If RPerl ever starts to be able to seamlessly e.g. mix in regexes, use $1, declare lexical vars without types, become able to use a string var as an integer without an explicit conversion, etc etc, then I will of course revise my opinion of it.