Speedup of the Archetypes reference engine.
Note: This code has been merged into Products.Archetypes and will be part of any Plone 4.1 and later release.
At the heart of the Archetypes reference engine is the reference_catalog. This is a set of catalog indexes used to perform the actual query lookups.
The choice of using a ZCatalog has lead to some data structures which aren’t suited for handling references.
This project tries to work around some of the short comings of the internal implementation of the reference engine without changing the public API or making any other changes to the stored data.
The source code can be found at: https://github.com/hannosch/experimental.atrefspeedup
If you encounter any issues, please contact hanno (at) jarn (dot) com.
1.0a6 - 2011-08-02
- Protect against invalid references pointing to non-existent objects. [hannosch]
1.0a5 - 2011-08-02
- Don’t use isdisjoint, as it’s not available on Python < 2.6. [hannosch]
1.0a4 - 2010-11-22
- Deal gracefully with missing uid catalog entries for references. [hannosch, stefan]
1.0a3 - 2010-11-05
- Speed up getReferences and getBackReferences of Referenceable objects. We no longer load the reference objects themselves but work directly with the information on the brains. [hannosch]
- Speed up getTargetObject and getSourceObject on reference objects. [hannosch]
1.0a2 - 2010-11-04
- Support passing in a list of relationships into the getReferences and getBackReferences methods. [hannosch]
1.0a1 - 2010-10-18
- Initial implementation. [hannosch]