Quoting Niclas Hedhman <[email protected]>:
Stan,
Sounds like a very ambitious implementation. So, IIUIC, you even have
indexing/query support against the the primary storage? Or is it
separated store/retrieve for performance?
Looking forward to take a closer look, and definitely news-worthy if
we can bring this to production quality level.
Currently it's only indexing/query. It does not use RDBMS for actual
primary storage.
If one wants to use it as primary storage, it is possible to construct
an entity based on information in tables - I think I might even some
code for that. And even if I don't, it should be possible to implement
though, because according to all logic, as long as transformation
"Qi4j Entity" => "Data in SQL Tables" is accurate enough for arbitrary
queries, the transformation should be two-way, ie also "Data in SQL
Tables" => "Qi4j Entity".
It's just a matter of performance - will it just be much faster to use
some other method for storing the entity (JDBM or single table in
another RDB schema to store JSONized entity state) than use complex
algorithm to construct entity based on various bits of information
scattered over many tables in RDBMS. Especially nested collections
cause trouble (and possible baldness ;) ) in this process.
Thanks for interest - it will most likely speed up my code to appear
in Qi4j codebase. ;)
_______________________________________________
qi4j-dev mailing list
[email protected]
http://lists.ops4j.org/mailman/listinfo/qi4j-dev
