[ZODB-Dev] Relational datas into ZODB

[Pierre-Yves Delens]

Bonjour,

My context : building sector specifications - ZODB plain filestorage
----------------------------------------------------------------------------
----------------------
a. Working on ZODB for categorized  and [sub]component datas, for building
sector specifications and quantities applications.
Several classifications are intensively used; they are of course hierarchic,
and lead us to create per Project Data sets modeled following these
hierarchies,in ZODB [sub]objects.

We need to port (or perhaps not.. ?) a 'mapping' database created for
maintaining the relationships between more or less concurrent
classifications, namely, the Table-Of-Content of several concurrent
Building-specifications sets. The data-model was, in the old SQL world, made
of 'ManyToMany' relations. For example, with topics from 2 sets SpecA and
SpecB :

SpecA.Topic 01.01.00 <EQUIVALENT> SpecB.Topic A.01
SpecA.Topic 01.01.00 <CONTAINS> SpecB.Topic A.01.01
SpecA.Topic 01.02 <CONTAINED BY> SpecB.Topic A.01

Relations are pretty similar to thesauri items associations(broader term,
narrower term....)
In SQL world we used to create a specific table as to 'objectify' the
relations, as they say.


b. I think to keep in ZODB plain filestorage (for sake of config
simplicity); however, as SqLite is around, I do not exclude to mix both
storages, if really necessary.

My questions :
---------------------
1. Are there good practices/examples in ZODB and OOP for modeling persistent
objects for this purpose ?
2. IndexedCatalog on ZODB seems to offerthe features sufficient for
implementing this; but is this the right way ?
    - and BTrees ?
    - and query - relation.py
3. Should such datas best remain in Sql storage ?

Thanks on forward
___________________________________________________
P-Y Delens, ing-arch. manager
LIENTERFACES - PY Delens, sprl
mail : py.delens at lienterfaces.be
web : www.lienterfaces.be
___________________________________________________