Piers Cawley <[EMAIL PROTECTED]> writes:
> Leopold Toetsch <[EMAIL PROTECTED]> writes:
>
> > Piers Cawley <[EMAIL PROTECTED]> wrote:
> >
> >> I'm not sure you can optimize it to a jump opcode when you're tail
> >> calling another function can you? You could be tailcalling into a
> >> closure so you'll need to use invoke to do the right thing with the
> >> lexical stack etc.
> >
> > Argh, yes. What about:
> >
> > If the subroutine is constructed inside the sub, parrot sees the class
> > and can decide at compile time to either jump or invoke.
> >
> > If the subroutine is passed in, a runtime check could be done:
> >
> > if sub.isa("Sub") goto do_jump
> > invoke
>
> Umm... not sure quite what you mean there. Don't forget that the sub
> could be got by doing:
>
> find_lex P0, 'sub_name'
One of my projects is to allow tail-calls in languages/scheme (Don't
assume anything in the near future; i currently work on the io-system
and its near and far friends). My plan for this is to use a new
emitting instruction _tail_call_function which does emit
find_lex P0, 'name'
invoke # this does not change P1
instead of
find_lex P0, 'name'
save P1
invokecc # this uses P1 as return address
restore P1 # invoke P1 from the function brings us back here
invoke P1 # this brings us back to our caller
which the current _call_function would emit
Thinking a little about it, even call/cc is very simple to implement
# checking of arguments
set P0, P5 # first argument is the function which will be called
set P5, P1 # first argument of the called function is cc
invoke # tail call
No jumps are needed in this code. Jump should get translated
jmp/branch -> invoke
jsr/bsr -> invokecc
ret -> invoke P1
bye
boe
--
Juergen Boemmels [EMAIL PROTECTED]
Fachbereich Physik Tel: ++49-(0)631-205-2817
Universitaet Kaiserslautern Fax: ++49-(0)631-205-3906
PGP Key fingerprint = 9F 56 54 3D 45 C1 32 6F 23 F6 C7 2F 85 93 DD 47
