Skip to content

Improve collection storage #107

@tzaeschke

Description

@tzaeschke

Currently, storage of collections, especially Map/Set, are not terribly efficient. One problem is that during deserilaization, the hashcode() function of all keys is called when they are added to a map/set.
This is partially done to avoid problems with schema compatibility between different Java versions.

Solution:

  • We could provide custom implementations. However these may have to be different for different Java versions, like Java 6, 7, 8...
  • We could try to use the Serialize/Externalize function, possibly by providing a 'Stream' that replaces references to FCOs withd OIDs. This may still cause problems if the serializableVersionUID changes.
  • We definitly should provide our own optimized implementations of SCO-HashMap/Set. We then could use these and replace Java-HashMap/Set on the fly, unless the user specifies otherwise.

The last options sounds like the best.

Metadata

Metadata

Assignees

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions