On Tue, 20 Aug 2002 17:29:58 GMT, Joel VanderWerf
<vjoel / PATH.Berkeley.EDU> wrote:

>Thaddeus L. Olczyk wrote:
>> On Mon, 19 Aug 2002 18:20:14 GMT, Joel VanderWerf
>> <vjoel / PATH.Berkeley.EDU> wrote:
>> 
>> 
>>>Thaddeus L. Olczyk wrote:
>>>
>>>
>>>>BTW I did not say it takes half an hour to start Ruby. I said it can
>>>>take half an hour to start Ruby and get to the point where I left off.
>>>>( Or to the last bug I fixed. )
>>>
>>>This in itself doesn't convince me you need a core. If you keep program 
>>>state in persistable objects, you can just dump them. Of course if you 
>>>use singletons or dynamically generated procs, you will have some 
>>>difficulty....
>> 
>> And you think persistance is instantaneous?
>
>I guess that's just my perspective. For me, simulations can take hours 
>to evolve, but the objects can be persisted in seconds. I can imagine 
>otherwise, though.
In simulations you have manymany objects which are constantly
destroyed and created ( unless you recycle objects, but even then
the recycling dominates ) that's in part why they take so long. Only
a small percantage of the total objects in existance over the lifetime
of a program need be persisted at any moment to restore the state.

In most programs the lifetime of an object is a signifcant percentage
( at least 10% ) of the program as a result at any time a large
percentage of total objects is present and need to be persisted.
Therefore the persistence and especially the restore ( where the
objects have to be recreated ) mechanisms take almost as long
as the Method of "run till you get to the desired point".