Dan Sugalski <[EMAIL PROTECTED]> wrote:
> Well... about that. It's actually easily doable with a bit of
> trickery. We can either:
I have trickery number 4) here. Dunno if its doable, but worth
considering IMHO:
Here is mmd.pasm (using bxor but substitute any math/bitwise/... op).
Comments inline.
_main:
new P16, .PerlInt
new P17, .PerlInt
new P18, .PerlInt
set P17, 0b101
set P18, 0b100
# might call a PASM sub or not, who knows
bxor P16, P17, P18
#
print P16
print "\n"
# plain MMD
# install PASM handler for overloaded bxor
.include "pmctypes.pasm"
.include "vtable_constants.pasm"
find_global P19, "PerlInt_bxor"
mmdvtregister .VTABLE_BXOR, .PerlInt, .PerlInt, P19
bxor P16, P17, P18
print P16
print "\n"
# So, now when the compiler see's[1] a mmdvtregister .. BXOR, it changes
# the emitted code sequence of all bxor_p_p_p opcodes to:
# recompile each bxor_p_p_p
mmdvtfind P20, .VTABLE_BXOR, .PerlInt, .PerlInt
isnull P20, bxor_normal_1
set P5, P17
set P6, P18
set P7, P16
set P0, P20
pushtopp
invokecc
poptopp
branch bxor_done_1
bxor_normal_1:
bxor P16, P17, P18
bxor_done_1:
# end bxor_p_p_p
# the recompiled code could be compacted a bit by some helper opcodes,
# but it shows what could happen
print P16
print "\n"
end
.pcc_sub PerlInt_bxor:
set I16, P5
set I17, P6
bxor I18, I16, I17
set P7, I18
invoke P1
[1] would need a flag in loaded byte code too or such
Comments?
leo
