Hello,

We have a scenario where we need to keep a bunch of objects in Cache with a structure like this (Adapter):

for example, tables such as:

client -> param (param_name_id, param_name_value) -> param_name (param_name_id, param_name)

So if we have a large ClientEntityCollection in memory:

client -> param (-> param_name "some param name 1") param (-> param_name "some param name 2") param (-> param_name "some param name 3")

client -> param (-> param_name "some param name 2") param (-> param_name "some param name 3")

etc.

There will be many copies of the same param name records throughout the EntityCollection. This is a simple example, but we are wondering if we can keep a quite large collection in the server cache. The root Entity in the collection will references many sub-entities used throughout the collection by other root-entities.

What are the memory implications?

Does LLBLGen keep separate copies of this data in memory or perhaps use references to avoid duplication of data?

client -> param (param_name_id, param_name_value) -> param_name (param_name_id, param_name)

So if we have a large ClientEntityCollection in memory:

client -> param (-> param_name "some param name 1") param (-> param_name "some param name 2") param (-> param_name "some param name 3")

client -> param (-> param_name "some param name 2") param (-> param_name "some param name 3")

The example is not clear for me.

But if you are asking about keeping unique instances in memory rather than duplicates. Then maybe you should be using a Context. Please refer to the LLBLGen Pro manual's section "Using the generated code -> Adapter/SelfServicing -> Using the context"

For example,

3 tables (project, client, client_contacts)

In code we have an EntityCollection of project objects:

projectA -> client1 -> client contacts for client 1 projectB -> client1 -> client contacts for client 1 projectC -> client1 -> client contacts for client 1 projectD -> client1 -> client contacts for client 1

There is a lot of duplicate data.

Does LLBLGen treat all of this as separate instances or references to save memory?

In our case we would prefer "references" to save memory as it will be used in a data cache type role.

Does LLBLGen treat all of this as separate instances or references to save memory?

Separate instances unless you use a Context.