Re: Heatbug.java continued..

[Miles Parker]

Miles T. Parker
Software Engineer
The Brookings Institution  1775 Mass. Ave. NW  Washington, DC  20036
http://www.brook.edu/es/dynamics/models/ascape
mailto:address@hidden  voice 202.797.6136  fax 202.797.6319

>>> Marcus G. Daniels <address@hidden> 01/13/00 08:53PM >>>
>>>>> "MP" == Miles Parker <address@hidden> writes:

MP> bestLocation = getHostCell().findMaximumWithin(maximizeFor, 1, true);

>>This is a good thing, I'm think, for a teaching example, but I wonder
to what extent that non-professional Java programmers (e.g. academic
researchers) ever feel a performance cost to implicit allocation like that
(and the cost of automatic reclamation of it).  

The code in fact does no actual object allocation except for locally scoped 
primitives. The code simply sends a reference to a location that already 
exists. Yes there are _reference_ allocations, but CMIIW I don't believe these 
are broadly speaking any more expensive then primitive 'reference' allocations. 
In fact, I ruthlessly excised any object allocation from the core engine after 
discovering what a huge hit object allocation is in Java.

The point about hiding details from non-programmers is an important one; see my 
responce to Benidikt. But the flip side of that is that someone who is 
concentrating on developing a framework may see opporunities for optimization 
that a user of the framework may miss or not have the time to exploit.

-Miles




                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.