Advantages of using language integrated queries technology.

Mitter, Florian ; Stopper, Markus]

Abstract: Current object oriented programming languages require additional application programming interfaces (API's) to access the stored data in relational databases. Therefore an investment of learning-overhead is necessary. Additionally missing compiler support and IntelliSense feature do not simplify the situation as well as the case for doubling-up possible error sources regarding to be on the way in two worlds. A combination of features in upcoming Microsoft C# 3.0 (generics, anonymous delegates, lambda expressions, query expressions--LINQ) can achieve the desired goals without facing these disadvantages. While in beta phase this solution lacks usability, but Microsoft company announced to implement LINQ in Visual Studio 2008 (Codename Orcas) with many more upgrades like a control element for an appropriate data source. Substitutional there are different object relational mappers available that image database tables in objects, but if the company's development progress will be as expected, after all LINQ technology would be the most effective solution for the creation of data driven applications.

Key words: language integrated query, generic query syntax, compile time syntax checking, .net framework technology.

1. INTRODUCTION

Language Integrated Query (LINQ) and in particular LINQ to SQL (DLINQ) reduces the complexity to access relational data by enabling direct data access in form of objects, attributes and methods. LINQ will be a native part of C# 3.0 and VB 9.0 in the next version of Visual Studio which development code name is Orcas (.NET FDC, 2007). It will be an integrated part of .NET Framework 3.0 and might get additional enhancements (like control elements etc.) in further patches and releases.

2. STANDARD OBJECT TO SQL CONVERSION

To run a database query out of a programming environment, an appropriate command string has to be assembled, which is then executed by specific methods of superior technologies (ODBC, DAO, OLE DB, ADO or ADO.NET). Each of them has its own object model, syntax and dependencies to other components and libraries (Sceppa, 2006). But they have in common, that the returned amount of data rows can be stepped through by a pointer (per next-function). To enhance performance for multiple use of the same query (in a loop) also state parameters can be specified. In this case, the query can be precompiled and only the parameter will be changed before every execution. But all in all, none of these ways is as flexible as it ought to be.

3. PROBLEMS OF CONVENTIONAL METHODS

The conventional methods of data access have their entrapments during the development process. In many cases the developer has to take care of problems that should have been handled by the supporting environment. These are simple problems like data type conversion and string concatenation on the one hand and lack of usability (no IntelliSense/Wizard) on the other hand. In brief, the developer needs to know about an extra technique and its shortcomings as well as the outcome are exposed unnecessary mistakes that could not be ignored. The former mentioned technologies should have been database independent. Due to this disadvantage all classic solutions simply implement a query string. But as soon as return values occur they have to be separated and stored in different variables of appropriate data types for further handling. The navigation through the received data rows has to be done by a pointer. This pointer (or cursor) has several very important attributes. For example a forward-only cursor is of course much faster than a dynamic cursor, but the programmer is subordinated to some restrictions that he has to keep in mind. In general, it is a trade off between performance and flexibility in application design. For a following update, the values need to get concatenated back into the right format. Editing persistent data out of a programming language therefore requires nearly as much additional effort as the underlying read/select operation. Grabbing persistent data this way is like an impurity in a programming language. This is why possible errors can not be spotted by the compiler before they happen during runtime.

4. APPLICATION AREA OF LINQ

Currently DLINQ is in beta phase and therefore it only supports MS SQL Server. While this deficit will get remedied, another one might still occur. LINQ does provide data manipulation language methods, but no form to run data definition language or data control language. Assuming that for the final version too, later changes on the database scheme have to be implemented manually in the already existing object model, bearing in mind that this might result in a complex task. Next to DLINQ, which is constricted to SQL server at the moment, there are related methods to access data out of XML (XLINQ) or even any object, entity or array that is able to fit into the IEnumerable data type. Like mentioned earlier LINQ is still in development. Components might take their time until they are bug-free and several wizards still would be necessary to improve usability (Channel 9 Forums, 2007).

5. LANGUAGE INTEGRATED QUERIES

[FIGURE 1 OMITTED]

DLINQ provides an SQL like syntax to select, group and sort data out of a relational database but also to add, alter and delete it. One entity class has to be created for every database table and this class has to contain a private attribute for each table column. The existing class "DataContext" provides an interface to a database management system which is assessed by a connection string. The emerging object can be used to access the database tables for reading or editing. Because LINQ is a native part of all .NET languages, it is able to provide IntelliSense. Additionally the compiler can assure the correctness of the query as the code compiles (ADT, 2007). Figure 2 shows a C# code example that runs against an array and figure 3 shows the result of it.

[FIGURE 4 OMITTED]

6. BLINQ

BLINQ is a tool that automatically generates ASP.NET websites based on an SQL server database. Such a website then contains pages to search, edit or insert data into each table of the underlying database. Additionally existing data rows can be deleted as long as relational integrity does not prohibit. BLINQ is mentioned in this paper because it uses LINQ just the way a programmer would do. The needed entity classes for the tables are built automatically and several ASP files are generated for each of them. Before BLINQ can be used, the programmer's environment has to be patched of course, since it is not included in any ordinary framework. After this is done, the connection string to a database is needed, and seconds later BLINQ constructs a whole linked website, mapping the database model. This website can easily be improved by login functionality or other useful features, and also the style sheet is easy to change. The database column names label the form fields of the front-end, so these labels should get renamed in Visual Studio. A bit of additional work appears when the database uses numeric values as primary keys instead of real data. In this case nearly every relation in a database model has to be altered in the website. But as long as the programmer knows where to go to, it will not take long. Figure 6 shows a foreign key field how it is generated by BLINQ offering the numeric value to select. Figure 7 shows what the programmer should make out of it. This was achieved by simply changing the DataTextField attribute of the drop-down-list to the column name that should be visible instead of the numeric value of the country. As can be seen in figure 7, the sorting inside a drop-down-list or a grid view is not attained yet. Therefore the programmer needs to dig into the code a bit more.

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

7. CONCLUSION AND FURTHER RESEARCH

LINQ raises the impression (further development assumed) to solve a big and wide spread problem: the paradigm change between relational and object model. Now that object oriented programming has already reached a stable point it is only a question of time when the according data access methods would get thought over (Stopper & Angerer, 2000). The more followers this new technology will get, the more time and cost will be invested by the programming industry and the usability as well as the performance of emerging competitor products will arise. Even now there exist object to relational mappers of other originators (e.g. Hibernate). The deficit of these is simply that they are not embedded in any programming language.

8. REFERENCES

.NET FDC, Framework Development Center (2007). The LINQ Project, Available from: http://msdn2.micosoft.com/enus/ netframework/aa904594.aspx Accessed: 2007-09-06

ADT--Application Development Trends (2007). Anders Hejlsberg on Project LINQ, Available from: http:// adtmag.com/article.aspx?id=20451 Accessed: 2007-09-06

Channel 9 Forums (2007). Chatting about LINQ and ADO.NET Entities, Available from: http://channel9.msdn.com/Show Post.aspx?PostID=202138#202138 Accessed: 2007-09-16

MS ASP.NET (2007). The sandbox projects--Blinq prototype, Available from: http://www.asp.net/downloads/sandbox/ blinq/ Accessed: 2007-09-16

Sceppa, D. (2006). Developers book--ADO.NET 2.0, Microsoft, ISBN 978-3-86063-541-4, Microsoft Press, Redmond, Washington

Stopper, M. & Angerer, B. (2000). An Automatic Streaming Service for Open Data Exchange in Factory Automation Systems using XML, Proceedings of the 4th International INES Conference, 17-19 Sept. 2000, Portoroz, Slovenia

Mitter, F[lorian] & Stopper, M[arkus] *

Fig. 2: C# LINQ example

using System;
using System.Query;
using System.Collections.Generic;

class app {
 static void Main() {
 string[] cities={"London", "Paris",
 "Madrid", "Berlin", "Vienna",
 "Vaduz", "Budapest", "Dublin"};

 IEnumerable<string> expr =
 from c in cities
 where c.Length == 5
 orderby s
 select s;
 foreach (string item in expr)
 Console.WriteLine(item);
 }
}

Fig. 3: Result of code shown in Fig. 2

Paris
Sofia
Vaduz

Fig. 5: BLINQ command line

blinq /t:c:\CapitalCities /server:localhost
/database:dbCapitalCities /user:usr1
/password:pwd1 /namespace:CapitalCities
/vDir:CapitalCities /pageDataSource