Cadec 2004
Vi var cirka 60 kunder och kollegor som lyssnade på fem intressanta presentationer om nya metoder och tekniker.
Här nedan hittar du presentationerna. Tveka inte att kontakta oss om du vill diskutera ämnena ytterligare med oss.
Presentationer
Johan Eltes
Large projects are forced to make many technical decisions; spanning from how to achieve a service oriented architecture to standards for CM and source code structure. Rapid decisions are often made as a result of intense workshops with whiteboards and markers. But what happens next? Far too often, the project manager has to report severe project delays due to technical refactorings of various kinds. It in not uncommon that as much as 30% of the initial resources of a project are devoted to discussing the plethora of architectural issues that sooner or later must be addressed. The Rational Unified Process and other process frameworks points to the Reference Architecture as an important tool to relieve the burden, enabling projects to focus on the business.
This talk presents experiences from both developing and implementing reference architectures in large, mission critical projects.
"JavaServer Faces - a comparison with other techniques"
Peter Norrhall
Some 10 years ago, the first RAD (Rapid Application Development) tools entered the scene, enabling us to rapidly develop both prototypes and complete Client/Server systems. By picking components from a tool palette and place them in a form, we could suddenly build applications in no time. But similar tools for building web based systems have not shown up, and the more successful tools and frameworks that do exist are not built on a common standard. This is what the JavaServer Faces standard (JSR-127) is supposed to change.
This talk will demonstrate the basis and goals of JavaServer Faces, and compare it to similar techniques.
"The Deployment Descriptors are gone - Metadata Attribute driven J2EE development"
Magnus Larsson
The complexity of J2EE with deployment descriptors and infra-structural interfaces (EJB) can sometimes be overwhelming for a development project. This complexity is usually addressed by using expensive enterprise development tools, or by large and complex modeling/code generation frameworks and tools. A promising, alternative approach which lately has gained momentum, is the so- called metadata attribute driven development. The hallmarks are efficiency and quality gained from simplicity, repetitiveness and control.
This talk will introduce the foundation of metadata attribute driven development, exemplify its usage and discuss how the technique will be introduced in the next version of both J2SE and J2EE. Examples from J2EE will be covered (based on XDoclet) and put into context as part of a development environment for both the small one person project to large-scale projects with several cooperating developers.
"Automated acceptance testing - a Holy Grail?"
Björn Beskow
With the rise of Agile Methodologies, Testing has received a central role as one of the critical success factors for agility. But Unit Testing, while tremendously popular and valuable, is not in itself sufficient. The interesting (and hence error-prone) aspects of most applications often take place between units, not inside them. This is where Acceptance Testing comes into play, opening a whole range of interesting questions:
- What is really the difference between Unit Testing and Acceptance Testing?
- How can Acceptance Tests be expressed in a language which makes sense both to customers and developers?
- Can Acceptance Tests be automated as cost-effectively as Unit Tests?
- Can we use some of the tools and lessons learned from Unit Testing in automating Acceptance Testing as well?
In this talk, principles, frameworks and best practices for automated acceptance testing are described.
"Is JDO (Java Data Objects) better than Entity Beans?"
Jan Västernäs
In the beginning, there were only objects. They were created, lived and communicated with each other. One day, someone turned off the power, and the need for persistency became obvious. Nowadays, there are no long-lived objects left, only a huge amount of loosely coupled DTOs flowing back and forth through our service oriented components. But the need for persistency of course remains. Since object databases haven't reached any substantial market penetration, we have to solve the mismatch that occurs when objects are to be stored in relational databases. There exists several different solutions to this problem, among them JDO which has received quite some attention and interest lately.
This talk will briefly describe how JDO works, and compare it to some alternative approaches.