On 7.08.2010 00:21, chromatic wrote:
On Friday 06 August 2010 at 13:23, Paul C wrote:
As an aside, is this doubling in size each time considered a good thing?
Seems okay up to about 64 or so, but then the table gets big fast. I'm sure
everyone knows the trick of using a vector of sizes and rep
On Sun, Aug 08, 2010 at 01:03:25PM -0500, Patrick R. Michaud wrote:
> On Sun, Jun 27, 2010 at 04:50:18PM -0700, Stefan O'Rear wrote:
> > > We may be able to add a register allocation step to POST.
> > >
> > > Ideas?
> >
> > A cheap half-way solution would be to add code to PAST::Compiler to null
The dynop_mapping branch is an attempt to deal with the dynop
numbering issue when loading PBCs that use different dynoplibs in
possibly different orders. It is currently in working condition,
albeit with a few snags:
* There is no longer a global table of opcodes, so OpLib needs
rethinking. Maybe
Patrick R. Michaud wrote:
> On Sun, Jun 27, 2010 at 04:50:18PM -0700, Stefan O'Rear wrote:
>> > We may be able to add a register allocation step to POST.
>> >
>> > Ideas?
>>
>> A cheap half-way solution would be to add code to PAST::Compiler to null
>> temporaries when they are no longer needed
