Memory-related failure condition with active dataset size in SPSS Statistics desktop client

Context

IBM SPSS Statistics desktop client can report memory-related failure conditions when an active dataset size exceeds available working storage. Behavior differs between 32-bit and 64-bit builds and between in-memory and disk-based processing modes. The Python and Java extensions run in separate integration processes that can retain memory across script executions. Documented practice advises explicit release of large objects and restarting the integration to avoid accumulation.

Design uncertainty

Memory usage for active datasets is managed internally with no public deterministic API to query current working memory per dataset. Users must infer usage from system monitors and documented size guidelines. Limits for .sav files and working storage are version-sensitive and differ between desktop and server deployments. The interaction between extension process lifecycle and host memory accounting is not fully specified.

What is the documented condition under which the desktop client reports a memory-related failure for a large active dataset in 64-bit versus 32-bit editions? Does the client provide a deterministic way to query working memory per active dataset without relying on external system monitors? Is restarting the Python or Java integration process considered sufficient to guarantee release of memory retained across successive extension script executions?