Archive for September, 2005
JAOO, Thursday 29th
Full day Tutorial: JDO, Craig Russel & Marc Prud’hommeaux
Good tutorial on a something that I am quite sure I will never use, probably not many other either. A bit sad since there seem to be quite some efforts put down into JDO2.0 soon to be released. I don’t really understand why Sun is pushing two all-in-all pretty similar persistence solutions, JDO and EJB3 … both obviously owing much of its design to Hibernate BTW.
Sun… it is high time to kill one of your babies. Don’t keep any of the two on technical merits. Instead consider user base and expected industry support.
No commentsJAOO, Wednesday 28th
-
Keynote: Tools for Software Architects and Future Directions in Modeling, Microsoft
Two guys from Microsoft showed what Visual Studio will look like in the future: a tight application containing more drag-and-drop programming than ever before. If this is a good thing, I don’t know? The guys from Microsoft gave a really tight presentation … but I am too bored right now to write anything about this. - Dynamically Typed Languages on the Java Platform, Gilad Bracha
Gilad discusses why it is hard to implement dynamically types within the JVM and what tradeoffs Sun are dealing with here. Interesting but quite far from my everyday consulting. On a side note: Gilad seems to be a clever guy. - Java Technology Performance Myths Exposed, Gil Tene & Ivan Posva
Gil and Ivan goes through 7 myths of hack:ish ways claimed to optimize performance in java. They tested those myths agains 7 JVM:s on different platforms which led to the quite unsurprising conclusion: Don’t try to optimize java on the local level. Most things that were said to be expensive (like try-catch and object instantiation for instance) are very rarely a problem. The cost of maintenance is most likely way higher.My take: Knuth said the very same things quite some years ago, and anyone trying to do the plain stupid optimizations described on this presentation I would without second thoughts kick out of my team no-matter the project.
- Are Java Fat Clients dead?, Adam Bien
Adam discusses the pros and cons of ultra-thin, thin and fat clients. As we all know the main reason for thin clients most often is ease of distribution. In the same time the problems are many: Duplication of validation logic on client and server side, broken domain model when converting it to a VO-model for client use to mention a few. We all have seen this. Adam lands the presentation in a push for the rich client together with a final…
…thin/fat-client should NOT be a management/marketing-decision. - EJB3 Persistence API, Marc Prud’hommeaux
Marc presented the coming persistence-API of EJB3. I conclude that this is (almost literally) Hibernate with annotations. Nevertheless interesting. - Monitoring and Managing in Java(TM) SE 5, Mandy Chung
Mandy showed JConsole, discussed the new monitoring features of JDK1.5 and discussed all pros and cons of MBeans, open types and MXBeans for plugging in custom configurability/monitoring in your applications. My conclusions from this presentation are two: 1) The monitoring possibilities on the JVM itself is something to put in the toolbox for the future. Acctually I believe I will have immediate use of it in my current project. 2) When writing own component the MXBeans seems to be the best approach.
JAOO, Tuesday 27th
-
Key-note: Beyond Agile – Smart. , Ivar Jacobson.
My impressions from this key-note is that Ivar is a grand vision man that in some aspects challenge the thoughts of the agile community. Generally I think that is good in this Agile-hallelujah-era. However, Ivar seems to be too much heavy-weight-process and top-down approach than I really sympathize with. -
Simple J2EE, Eberhard Wolff
Eberhard elaborated around the simplifications of J2EE in the upcoming release. Cannot have made much of an impression on me since I cannot remember a single thing from this presentation. Weird. -
P2P – new architecture ideas with JXTA for J2EE, Adam Bien
Adam presented an alternative to the common service lookup in enterprise applications. In short: By using a peer-2-peer network for service discovery a better redundancy, fail-over can be achieved. Interesting but I am not really sure the approach carries its own weight. -
Trends and the Future of Enterprise Java, Floyd Marinescu
A look into “important techniques for the future” with Server Side father Floyd gave:- AOP (3%)
- Dependency Injection (30%)
- JDK Annotation (20%)
- Java as a platform for scripting languages
- AJAX
- Webservices/SOA
- JSR 170 – Content repository
- EJB3 might make EJB cool again
(the figures in the braces approx. represent how many in the audience that said they had been using the techniques in a real world project … I am bit surprised, thought it would be more than that.)
This was one of the best presentations on JAOO2005.
-
JUTLAND: Java Unit Testing – Light, Adaptable ‘n’ Discreet, Kevlin Henney
Kevlin presented an alternative testing framework, called JUTLAND, which offered better solutions for some of the shortcomings of JUnit:- Providing pluggable assertions.
- Better exception (i.e. negative) testing.
- Facilitating test-object reuse.
- Being less intrusive.
Although I doubt we will ever be using JUTLAND as an alternative to JUnit the presentation was nevertheless interesting … in pointing out the legacy the JUnit in its current version actually carries. Having modified JUnit myself, and having seen JUnit under the hood, I really see a point in getting this discussion going. There must be better and more modern ways to do unit testing.
JAOO, Monday 26th
- Key-note: The Zen of Free – The Virtuous Cycle of Open Source, Simon Phipps / Sun Microsystems.
Simon says (more or less): With “closed door software development” you must understand that most of the smartest people in the world are outside that door. He discusses the benefits of open that door and specifically stress the asymmetry in the open source feedback loop – i.e. the fact that only the best code gets submitted in to the open source projects from a wide user base. Also he points out that there is and will be room for businesses capitalizing on adding value to open source products only as long as that delta (of value) is positive.
…and sums up with the following:
o Participation age is here – inevitable and ongoing
o Open source is the methodology of the partition age, sort of a connected capitalism.
o Open source is not enough. Standards are still needed. - The Joy of scripting, Dave Thomas.
Dave Thomas sets the stage about scripting. Nothing much to say about that really. - Ruby on Rails – Part One, David Heinemeir Hansson
David very rapidly created a very interesting demo application in front of a very impressed (as far as I could tell from the reactions) audience on the rails framework. I was looking on ruby some years ago but haven’t been following what is happening with it until just recently. This presentation (together with an equally interesting rails demo made by a colleague back at Citerus) now really reminds me that I will have to look deeper into this as soon as possible. The reason is simple: I think I need the best tools possible in my toolbox as a system developer. Ruby on rails might well be it. - Ruby on Rails – Part Two, David Heinemeir Hansson
This was a traditional presentation on the rails framework. A very selling presentation too. I won’t go into details. It suffices to say that David is a good and confident speaker and gave most of us a real eye opener. - Pushing the limits, Klaus Marquardt
A very boring presentation taking a meta view on patterns. - Past, Present and future of patterns, Frank Buschmann & Kevlin Henney
Equally boring (almost).Honestly… please relieve us of more patterns discussions on JAOO the coming years. Patterns in themselves are good and in many ways very useful, but come on … the “news value” of these two presentations were almost zero.
JAOO, Sunday 25th
Full day tutorial: Scaling Agile Development with Robert C. Martin and Dave Thomas.
While being very entertaining – given the skill of the speakers – this tutorial nevertheless seemed to have a quite misleading title. Basically Robert explained most of the agile practices whereafter Thomas shared his experiences of a scaled up agile project, the Eclipse platform. Unfortunately I can’t say I left the tutorial with great insights on how to really scale up agile development other than that it seems to be hard and will require a hierarchical structure (scrum of scrums) and some upfront planning. Still, as a remainder of the XP-practices and their foundation the tutorial had its merits.
No comments
This blog is written by me, Tobias Hill.