On Tuesday 01 August 2006 02:52, Leopold Toetsch wrote: > Two things come to my mind: > 1) why is it creating 2 interpreters?
Per my experiments, this worked out the best. That is, if there's an error in the compiled code, reusing an interpreter gave weird answers. I haven't tracked this down in more detail. > What is the $parent used for? It's the parent of the interpreter created in the pir_output_*() functions. Creating new subsequent interpreters without providing $parent gave weird encoding assertion errors. > And related: is $interp ever cleaned up by calling Parrot_exit()? Nope, but I'll put a destructor in Embed::Parrot::Interpreter when I get that far. > 2) This looks a bit bogus (there is no 'V' signature char): > > my $out_pmc = Parrot::Embed::call_sub( $interp, $sub, 'PV', '' ); > my $output = Parrot::Embed::get_string_from_pmc($interp, $out_pmc); > > It could just be: > > my $output = Parrot::Embed::call_sub( $interp, $sub, 'Sv'); > > by returning the STRING* directly. True, but I prefer to use the PMC methods rather than fiddling around converting strings to and from the C format. If this API changes to send and receive char * instead of STRING *, I'll happily use that. -- c
