hdolder.com srl

Specifications executed directly on a virtual computer 

Operational principles of M&P

As we will observe in the course of this document, M&P is a generalization of the DBMS (Database management system) concept.

M&P applies the operational principles of the DBMS at a higher level, the full applications’ level. 

Level-wise, M&P is for an application what DBMS is to a database. By analogy, we may say that M&P is an AMS (Application management system).

M&P generalizes and extends the most used concepts in the DBMS such as Data Model (Schema), Modelling Language, Meta-data repository, Rule enforcement, Automated optimization, Procedures, Triggers, etc.

The Management System role of a DBMS is performed by M&P‘s Virtual Computer (MPVC).

Modelling language

M&P uses XML as its modelling Language and as meta-data (model) repository also.

Procedures and triggers

The triggers and procedures of M&P are expressed in mpLISP. 

mpLISP is a minimum LISP dialect specifically defined for this function. 

The M&P‘s virtual computer integrates an "Application Runtime Library" (ARL) that provides top level functions used by M&P applications.

A key high performance component of the virtual computer is the LISP interpreter, which interprets the mpLISP code.

The functionality of the LISP interpreter is easily extensible using any of the MS .NET Framework available languages (over 30), especially C# and Visual Basic.

This extensibility of the LISP interpreter is of great importance because the main benefit is that the coding can be done in any .NET Framework language and not in mpLISP. 

The mpLISP code acts as the “connective tissue” assembling the system’s components and improving their interoperability.

This also makes easier the implementation of distributed applications, which run on several processors, as well as the flow of data and code.

Automated optimization

Similarly to the DBMS, the virtual computer provides an "automated optimization" of the applications that allows the achievement of higher performances than the applications developed with all other current technologies.

M&P offers a Framework where the model and code of the application are "hosted".

Plug-and-play components

M&P’s Framework implements plug-and-play components named plug-ins (a plug-and-play component is a component that can be incorporated at run-time without the need to reference it at compile-Time).

The code for the application (LISP interpreter’s extensions) is hosted in one or more plug-ins.

The plug-ins, which must be used in the different runs, are specified in "Starter-Files". 

A plug-in can also incorporate other plug-ins at run-time. This advantage allows the implementation of systems with a high degree of "Dynamic Self-Organization".

A M&P Application is defined by means of the following elements:

• One or more models·        

• One or more plug-ins         

• One or more starter-files 

M&P‘s virtual computer interprets these elements at Run-Time.

 

Extensions to the DBMS concept

Use of the code as data

The main data-structure of the LISP language is the list (LISP = LISt Processing) but LISP code is also expressed as a list. 

This "isomorphism" of data and code allows the code to be treated as data in one occasion and executed at a different moment.
This way the code can be stored in a database, file, or transferred to a remote system, in the case of distributed applications, for execution.

The code can be also created or modified “on-the-fly“ through the available list processing functions.

In-Memory-Data-Base (IMDB)

An In-memory-Data-Base (IMDB) is a relational database hosted in memory, that lets you eliminate (or minimize) disk access by handling and storing data within the main system memory.

M&P operates with two database levels:

• In memory (IMDB)         

• In disk (DBMS Database)

and deals with high-level operators (though simple) for the data transfer between both levels. M&P establishes an implicit "mapping" between the two levels allowing one model to describe the scheme of both levels.

Dealing with data in memory simplifies the coding and maximizes application performance.

Within M&P, an In-Memory-Data-Base is created using only two types of data structures:  Entity-Set and Instance-Collection. 

An entity-set defines an IMDB which contains one or more instance-collection (as a first approach we may compare them to relational tables).

Independent instance-collections can also be created.

An M&P application operates using entity-sets, instance-collections, and lists. 

Operators that use the definitions specified within the model create the empty entity-sets and instance-collections on demand.

Later, a group of high-level operators enable the loading of data from the DBMS to the instance-collections and vice versa. 

The data transfer is Transactional and is subject to a Concurrency Control (M&P uses "optimistic concurrency").

The virtual computer manage automatically the DBMS connections and generates on-the-fly the SQL sentences required using the definitions in the model. 

In M&P, the data Load and Update operators are "SQL-less" though if convenient, SQL sentences may be used.

M&P’s IMDB allows data "browsing" in a simple way (also by means of high-level operators), at a low cost and with top performance.

User interaction

User interaction is accomplished through SKINS ("ultra-light" forms with no logic within). The Controls in a SKIN connect directly to the Entity-Sets and Instance-Collections with a Bidirectional DataBinding device (modifying the control’s content, the content of the connected data structures is automatically modified and vice versa). 

Rule enforcement

The virtual computer does the “Rule enforcement” preventing invalid operations. For this, the centralized logic found in the model and the plug-ins is utilized. 

For a DBMS, triggers are procedural code automatically executed as a response to a certain event occurred within the tables of the databases. 

The invoked procedure may restrict the access to certain data, stop its modification, record the changes for auditing purposes, etc.

M&P provides similar triggers, in the In-Memory-Data-Bases. 

The IMDB dynamically created on demand, based on a model, automatically invokes the indicated mpLISP methods from the model for all events concerned.

Through this mechanism is possible to validate and enable o reject  proposed changes for the data stored in the IMBD.

In M&P, the equivalent of  stored procedures and user-defined functions (UDFs) of the DBMS are implemented through mpLISP methods.

Model-driven system

DBMS are "schema-driven Systems" because they actively use schemas to conduct its operation. Analogously, M&P is a "model-driven System" because the system uses the models actively to direct its operation. 

Object-relational models

"While the relational model has been extremely effective in the last few decades, it's a model that targets a level of abstraction that is often not appropriate for modelling most business applications created using modern development environments."
Microsoft Corporation [R1].

M&P applications are based on relational databases.

To the present time, M&P operates indistinctly with four different RDBMS:

•  MS SQL         

• MS SQL CE       

• Oracle         

• MS Access Jet Engine

The same model interacts with any of these by changing the parameter in the model’s ConnectionString.

M&P adds to the RDBMS the functionality of object-oriented databases, transforming relational databases into Object-Relational Databases.

M&P incorporates the following elements of the object-orientation into its models:

OO Elements	M&P designation
Class	Entity
Object	Instance or Entity-Instance
Property	Attribute
Method [1]	Method
Inheritance [2]	Inheritance
Polymorphism	Polymorphism
Object Reference	Reference

[1] of Classes and Objects
[2] Multiple inheritance

Consequently, an M&P model is an OBJECT MODEL or an application BUSINESS OBJECT MODEL and its programming interface (API) is an object-oriented interface.

Object-oriented programming grants a great notational economy for the models, allowing very compact models as well as a better business logic structure. 

When faced with complex business logic the principles of encapsulation, inheritance and polymorphism that the object-oriented approach have, allows the designer to manage the complexity with great ease.

Implicit object-relational mapping

Object-Relational Mapping

In M&P, each Entity relates to a table within the database. Each Instance (or Entity-Instance) relates to a tuple within the corresponding table.

Broad-spectrum modelling

Object-oriented modelling lets a fast and simple description of Data-driven and Event-driven applications as well as  Process-driven applications often called "Workflow" applications.

Normalized databases

One of the great advantages of the object based modelling of M&P is that produces normalized databases directly and naturally.

Design and development

At the beginning, we submitted that M&P represents to the applications what a DBMS is to databases. Designing an M&P application is similar in large degree to designing a database. Therefore, the experience gained from database design is applicable to the M&P application design.

The increased productivity from developing applications in M&P lies in the following key elements:

• A compact, complete modelling system without redundancies that has a high calibre expressive power (notational economy)  
• An optimal mix between declarative and procedural programming· 
• The two basic structures that integrate in a database hosted in the main memory. 
• The implicit object-relational mapping· 
• The Bidirectional databinding between SKINS and IMDBs· 
• The high-level operators to write the procedure’s code 
• The Framework that provides an standard infrastructure for the applications· 
• The fast and simple modularization provided by the plug-and-play components· 
• The Model Player that provides a graphic navigational interface on the database data. This largely satisfies the user’s demand of information. It also allows the editing of database information subject to the virtual computer’s Rule Enforcement· 
• The model-driven SQL Query Builder facilitating the formulation of unexpected queries
• The virtual computer that interprets and process’ the models with high performance, adapting automatically to changes introduced in the models

The eBook describes also the following elements that yield high productivity:

• A simple generic mechanism, with a standard graphical user interface for Concurrence Conflict Resolution during the disk recording of the database update· 
• The Schema Evolution Manager that simplifies the change of the database schema when changes to the models are made
• The AUTOKEY identifiers that provide Universal Unique Identifiers, greatly simplifying the applications

Direct execution of specifications

The database designer provides the DBMS minimal information in the form of schemas and compact code for triggers and procedures, then the DBMS provides the necessary functionality for the interpretation and optimized execution of them.

A DBMS directly executes the minimal specifications of the database.

Analogously M&P directly executes the minimal specifications of the applications.


References

·         [R1] The ADO.NET Entity Framework Overview 

---

EXECUTIVE SUMMARY           [Español]