[elephant-devel] Elephant Collections

[Ben]

the elephant collections are not pulled entirely into memory.  IIRC it
tries to be as lazy as possible as far as loading entries.  you may
have to load a bunch of stuff if you're doing complicated joins.

the same goes for references to other persistant objects.  in general
they are stored in slots by ID and are loaded on-demand.  this is not
true if the pointed-at objects are not persistent -- e.g. don't use
the elephant metaclass.

B

On 1/5/06, Waldo Rubinstein <waldo at trianet.net> wrote:
> Hi again,
>
> I'm interested in learning about the collections supported by Elephant.
>
> I have a need to store [very large] "flat" collections. For example,
> I want to store 2 million objects of approximately 1K bytes in size,
> ordered by date (an object attribute).
>
> The only reference to collections in Elephant I have seen is the
> tutorial using btrees and indexed-btrees. What will happen when I
> load this many objects in a btree. When I get a hold of the
> persistent btree, will it keep everything in memory or will it read
> in memory the objects that I fetch from it?
>
> Also, I have another case. I have to store about 100 objects where
> each of these objects has a reference to a collection of other
> objects. The sub-collection of objects don't need to be in any
> particular sort sequence. I was thinking of using a resizable vector.
> However, I was wondering the same questions as above. If I fetch one
> of the 100 objects, will it automatically load in memory all the
> objects in its sub-collection? The sub-collection objects may be
> 30,000 elements of approximately 1K bytes in size as well. How will
> it behave if I need to traverse the sub-collection of objects of a
> particular object? Will it, again, load all the elements in memory or
> will it read from DB as needed?
>
> Thanks,
> Waldo
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel
>