On 5-Jan-06, at 3:20 PM, John M McIntosh wrote:
>
> Two changes, the one being reworking the allocation calls to allow  
> for small and normal object allocation without initializing to  
> nilObj or 0, or some value. This allows calls that say create  
> floats to get back an allocated object, then fill in with the data.
IIRC I put that into VMMaker quite a while ago. At Interval I had aVM  
variant that would create large objects without initialising and set  
the relevant header flags so that GC wouldn't get upset. The #new  
methods would then fill in the nils or 0s later before transmuting  
the new objecct to the right class in full. Although this takes  
longer overall it made a big difference in the process switching  
latency that was terribly important to us at the time.

The more likely beneficial answer to slow opening of morphic  
explorers (and just about any other damn morph come to that) is to  
improve the algorithm. That almost always wins.


tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
On a clear disk you can seek forever...