Sunday, August 26, 2007

project proposal

The scaffold for this proposal is taken from from the UOS website.

Purpose (problem statement):
To simplify the process whereby student's find clubs and societies that they are interested in. That is creating an new channel of communication between students and clubs and societies of the university.

Currently the only techniques in place to join students with the clubs and societies they are interested with is during O - Week in the week 0 of semester 1(and to a lesser extent Re - O Week week 1 of semester 2). Also there is a directory of all clubs that are registered on the Union website. Now the problem with this is that is hard to navigate, and is unsearchable. This web site does not provide a mechanism to formulate links between the students and the clubs.
By providing a new way to find the clubs and socs that students like it will help keep memberships in the clubs growing and allow more student's to have the complete university experience. The Clubs and Socs program provides an important part of the university experience.

Users / Actors:
The actors that will use the application are as follows:
1) Clubs and Socs Executive: It is their role to update and upload information about their particular club.
2) Clubs and Socs Administrators: These are the union execs that run clubs and socs, it's their job to add club and socs execs as part of the registration every year.
3) Students: who can search the system and get in contact with the clubs and socs they want.

Use Cases:
1. Add / Edit / Remove Club and Soc Exec
2. Add Club Information
3. Search for a Club
4. Contact a club (this one may change submit interest)

These use cases may evolve over time.

Thursday, August 23, 2007

recent labs + hibernate

Monday's lab for me well was out... epilepsy and all ... well i'll try to get the hibernate part done asap, also this weekend i will finish readings and write the description for the project.

Wednesday, August 8, 2007

extra lab

So I went into the lab today and did some work, i finished of the Ant JUnit part. Also i did the form data. The form part was quite easy however it was something that i had not used. Everything was integrated with Spring. The main issue that was quite new was the integrated with tags. The tagging library allowed the inbuilt Spring framework to be used at all areas of Web IO. The reason this is powerful is that we can then use the internal Spring functions to do things like:
  • Validation
  • DB connections
  • Log in etc.
I think this is much better than using simple standard xHTML tags. We can then form better automation and the system will be scalable.

Other things that I needed to mention that were important are:
  • Tagging is a critical part of Spring and i need to look at this a bit more.
  • JSP relies heavily xml and the web xml files. The strucutre of these needs to be learnt instead of relying too heavily on the files that are already there

Monday, August 6, 2007

lab 2

Lab 2 was not as productive since i ran into major issues with the JUnit part. This was because of a faculty install, and so i was the Guinea Pig for the solution. At the end of the day it did work and the JUnit tests (which i knew how to do) worked.
One thing i did learn though was the way to make ant do them for you (even though this took forever to get working) This can be useful in the long run. I will be going back to the lab on Wednesday to finish of section 3 and 4. This should be more productive as i will begin using things like tag libraries and forms, which i am new to.

Croosing chasms

Crossing Chasms: a pattern language for OO application frameworks.
Brown, K. and Whitenack, B.
Vlissides, J et al (eds), Pattern Languages of program design.

This paper deals with many issues with relation mappings to OO systems. This is a major issue in current real time OO centric systems. The paper concentrates on static mappings, that is how the objects map to tables and vice versa.
many of the issues raised come from basic DB theory and going from first Backus Normal form etc. Allot of the paper deals with the best way to deal with relationships and map these things from a design standpoint. Allot of this is hidden is we use something like Hibernate.
However we will return to this paper after we do some more things on Hibernate. However some key points to note:
  • Look at the way objects map to the DB early so this mapping has some significant effect.
  • Ensure you spend time with the database and map the relationships like you would when you going from 1st order through to 4th order BN.
  • You will need to give every object some unique key, in order to make the relational world happy, that is everything will need some form of primary key, and this will help to amp relationships (from simple 1 : 1 to inheritance mappings).
  • You will need to have some idea about the way the database stores primitive types so you know what format things will map to.
  • When dealing with the issues of relationships you should use direct access (i.e. pointers within the object) rather than relying on the presence of foreign keys which is inherently slow. However ensure that these do exist and are updated to ensure that when an object needs to be instantiated it can be.

Frameworks = components + patterns

Johnson, Ralph E
Communications of the ACM. Vol 40, (10) pp. 39 - 42

In order to understand frameworks you must first try and define them. We can define a framework in a computer sense as a reusable design of parts of a system represented by abstract classes. It provides a customizable interface for developers to use.
Frameworks are like components (such as those components in an OLE library) however they are more complex and also you can customise them allot further. A component can be thought as simply code reuse. A framework is more than that, it is actually design use as well.
Frameworks reuse design aspects like a template or a high level schema. The difference is that a framework has been described in a fixed programming language (Java or C++).
Reusing design has its benefits and its own cots. In the long run you can save time and money with Frameworks, however there is always the steep learning curve associated with developing it.
A pattern is a way to describe a solution to a problem in a certain context. Frameworks implement patterns in a specific way. That is they transform context specific ideas into some tangible code that solves a certain problem. For instance MVC is really a pattern, however the way Spring implements MVC can be though more as a framework.
Frameworks are in the middle between pure code reuse or design reuse. The allow us to use design principles and patterns in a language (and hence domain) specific way.