On Oct 9, 2005, at 10:29 AM, Oleg Broytmann wrote:

> Hello! Thank you for replying!
>
> On Sun, Oct 09, 2005 at 10:02:05AM -0400, David Binger wrote:
>
>> Are you talking about iterating over the keys of an index, and
>> loading
>> some subset of the values based on information that is encoded in
>> the keys?
>>
>
>    Yes, exactly. I'd like to write a program that will do all kind of
> searches through a huge list of objects, and I need to make the
> program
> speed- and memory-efficient. Iterating over indices seems like a
> good (if
> not the only) approach.

The volume of data that is loaded when you load an instance can be
managed
by moving bigger (non-key-ish)  data into sub-instances.    For
representing email
messages, for instance, you might keep subject and sender in a
persistent instance
with all of the other data in a sub-instance.  This is essentially
the same thing
as having separate information-rich keys, but it may be easier to
manage.
>
>
>> References are really the dominant structural feature.
>>
>
>    What is a "reference"? You don't mean deep object hierarchies
> (a.b.c.d),
> do you?

I mean a Persistent instance has a direct reference to another
Persistant instance,
without going through any application-level  index.  If the situation
requires following
chains of several references to get to the object of interest, then
that is indeed what
I mean.










>
> Oleg.
> --
>      Oleg Broytmann            http://phd.pp.ru/
> phd@phd.pp.ru
>            Programmers don't die, they just GOSUB without RETURN.
> _______________________________________________
> Durus-users mailing list
> Durus-users@mems-exchange.org
> http://mail.mems-exchange.org/mailman/listinfo/durus-users
>