Data Persistence Framework

To expand further on a rapid web application framework, we also need a complementary data storage mechanism.

In today's world, most data is persisted in RDBMS's like Oracle, Sql Server, DB2, Sybase etc. In my opinion most of the applications do not need to have dedicated schema for storing objects.

Instead they can make use of a tuple based design for storing various object instances. A tuple and its related operations can be defined succinctly using a Xml grammar (like given below:).

<AppObject1>
  <Meta>
    <Attribute1 key="yes" type="int"/>
    <Attribute2 key="yes" type="string"/>
    <Attribute3 type="datetime"/>
  </Meta>
  <Upsert name="CreateEntity">
    <Attribute1>@Attribute1</Attribute1>
    <Attribute2>@Attribute2</Attribute2>
    <Attribute3>@@getdate()</Attribute3>
  </Upsert>
  <Select name="GetEntities">
    <Attribute1/>
    <Attribute2/>
    <Attribute3>@@getdate()</Attribute3>
  </Select>
</AppObject1>

This Xml format or notation can be generalized across a whole class of problems. It fits the relational model quite nicely too. For example, one can imagine a database table having the following schema:

 0.  AppObjectId   BIGINT
 1.  AppObjectName NCHAR(50)
 2.  Type          NCHAR(50)  
 3.  SubType       NCHAR(50)
 4.  Bigint0       BIGINT 
 ...
 13. Bigint9       BIGINT
 14. Decimal0      NUMERIC(18,6)
 ...
 23. Decimal9      NUMERIC(18,6)
 24. String0       NVARCHAR(2000)
 ...
 33. String9       NVARCHAR(2000)
 34. TextDataId    BIGINT
 35. CreatedBy     BIGINT
 36. CreatedDate   DateTime
 37. UpdatedBy     BIGINT
 38. UpdatedDate   DateTime