On Mon, Dec 7, 2009 at 4:14 AM, chromatic <[email protected]> wrote:
> A good optimization step is enabling the use of compile-time constant PMCs in
> bytecode. Consider this code generated by NQP-rx:
>
> .namespace []
> .sub "fib" :subid("11_1259711141.42547") :outer("10_1259711141.42547")
> .param pmc param_16
> .annotate "line", 3
> new $P15, 'ExceptionHandler'
> set_addr $P15, control_14
> $P15."handle_types"(58)
> push_eh $P15
> ...
> .end
>
> The ExceptionHandler in $P15 is (effectively) constant and won't change
> throughout the invocation of the program. Recreating it (and calling a
> *method* on it to set the single type of exception it will handle) is busy
> work all but the first time. (This code comes from the NQP-rx fib.pir
> benchmark; we can get an easy 45% performance improvement from avoiding re-
> creating this EH on every sub invocation.)
>
> We should be able to create a single PMC once, at compile time, and refer to
> it throughout the rest of the PIR. To accomplish this, we need:
>
> * PIR syntax to declare a constant PMC and initialize it at compile
> time
> * PIR syntax to refer to such a constant PMC
For a non-EH, you can do something like:
.sub 'string_options' :anon :immediate
.local pmc opts
opts = split ' ', 'bytelength compare equal first index is last
length map match range repeat replace reverse tolower toupper totitle
trim trimleft trimright wordend wordstart'
.return(opts)
.end
.sub foo
....
.const 'Sub' options = 'string_options'
.end
> * tests for both
> * (possibly) a way to refer to local symbols in other subs, such as
> the label
> used with the set_addr op
This is the only problem with the .const syntax above.
> Arguably, we could instead use a way to associate an EH with a sub such that
> entering the sub always activates that EH, but then we have to consider access
> to the sub's local symbols within the EH, and that's trickier.
>
> -- c
> _______________________________________________
> http://lists.parrot.org/mailman/listinfo/parrot-dev
>
--
Will "Coke" Coleda
_______________________________________________
http://lists.parrot.org/mailman/listinfo/parrot-dev
