Speedo TODO List
- Bug to fix: The persistency by attachement must be checked at
commit
time too and can be able to forget attached instances which are no more
attached to a persistent instance. That implies to maintain in the
working a list of really persistent class and a list of persistent
class by attadchement. At the commit time Speedo must checks if among
the persistent class by attachement, some instances are no more
attached to a real persistent class.
- Horizontal mapping for the inheritance (Olivier is working on)
- Medor/query must support horizontal mapping for navigation.
- constant hidden field
- Vertical mapping for the inheritance
- Cascade delete
- JDO 2: interfaces
- JDO 2: use standard mapping
- JDO 2: query result type
- JDO 2: SUM, MAX, AVG in queries
- JDO 2: distinct in queries
- JDO 2: queries limit
- JDO 2: indexes generation
- Generate SQL script for database initialization (table creation,
foreign key constraint declaration, index declaration) TODO in JORM.
- Provide particular implementation of Collection and Set in order
to avoid the loading in case of very large collection. These
implementations would implement the collection/set methods with SQL
order (add ==> INSERT, remove ==> DELETE, contains ==> SELECT,
iterator ==> query, ...). The implementation idea is to consider
collection elements like persistent objects. The identifier would be
composed of the collection identifier and the element identifier (PName
or primitive value).
- Attribute a weight to the compiled query. This weight would be
based on the time to compile the query. Indeed an heavy query should be
kept in cache in place of light weight query.
- Distributed concurrency manager (use existing Perseus component)
- Distributed cache
- When prefetching an extent of an inheritance graph, with auto
references, avoid to load these auto references form the database if
already present in the prefetch buffer (but not yet in the map [PName
-> RowId]).
- Administration via JMX using Fractal
Back to the Speedo documentation