Data loses when reusing persistent objects

[polsm91]

If we assign two StorageObj to an in-memory StorageDict, and then call make_persistent their data is overlapped. The storage dict value name is used to generate the UUIDs, therefore, all entries get the same name and same UUID. They clash and overwrite each other.

In [2]: from hecuba import StorageObj

In [3]: from hecuba import StorageDict
    ...: class SObj(StorageObj):
    ...:     '''
    ...:     @ClassField none int
    ...:     '''
    ...:     
    ...: class Test(StorageDict):
    ...:     '''
    ...:     @TypeSpec dict<<lat:double>,val:__main__.SObj>
    ...:     '''


In [7]: a = SObj()

In [8]: a.none = 123

In [9]: b = SObj()

In [10]: b.none=256

In [15]: c = Test()

In [17]: c[0]=a

In [18]: c[1]=b

In [19]: c.make_persistent('name')

In Cassandra we have:

cqlsh> SELECT * FROM my_app.name;

 lat | val
-----+--------------------------------------
   1 | 66aaef29-4046-3827-9710-37073d04dda0
   0 | 66aaef29-4046-3827-9710-37073d04dda0

(2 rows)

cqlsh> SELECT * FROM my_app.name_val ;

 storage_id                           | none
--------------------------------------+------
 66aaef29-4046-3827-9710-37073d04dda0 |  256

(1 rows)