aLogcat market barcode
There are several “log view” applications on the market. All of them provide a means to send your log file contents, typically via email. This is one good approach, but it doesn’t handle the use case where you don’t have immediate access to a PC email client to view the results. aLogcat is an Android application that allows you to view your Android device log from the device itself. It provides a scrolling, color-coded log that is filterable by keyword and log level. It also supports output in various log formats. aLogcat also covers the send log use case by allowing a snapshot of the log to be sent off to another device.
This is mainly an app for developers, but it is also useful for power users that are willing to get involved with developers to help them find problems in their applications. Most Android developers are small scale hobbyists and can’t devote full-time effort and money to rigorous testing across multiple devices. I hope this app lowers the barrier for involvement of the average user in the development cycle.
After working on two large-scale, cross contains / platform, web application projects, I’ve come to the conclusion that Java EE is fatally flawed.
If someone asked you to state the most important, guiding principle of the Java language, what would you say? You would would probably say “write once, run anywhere.” Indeed, for Java SE this holds true most all of the time. In Java EE however, it is hopelessly broken. The reality is that every web container is slightly different in ways that force developers work around problems, avoid certain features, or even have different code paths. Different code paths in a language that was never designed for it.
Different aspects of Java EE have different levels of stability. JSP works almost the same across all containers. Things like JSF and Java Persistence (JPA) however are hopeless. My most recent experience is writing a web administration console using JSF for the OpenSSO project. OpenSSO claims to support at at least 8 different web containers. Every container required JSF tweaking that ranged from annoying to just plain terrible. Needless to say I had egg on my face to some degree over the choice to use JSF. It was a no-brainer for me. JSF is the chosen MVC framework for Java EE. My mistake.
The only container where JSF “just worked” was Tomcat. Tomcat isn’t a Java EE container at all.Tomcat doesn’t include JSF or any of it’s dependencies. We might be on to something here. Let’s look at another example: Spring. Spring is essentially a lighter-weight, simpler replacement for Java EE. Why is Spring so popular? One reason is that it is an elegant design. But another, perhaps more important reasons is that no matter what container you are developing on, Spring is Spring. It’s the same JAR, the same code, the same implementation. It makes very few assumptions about the capabilities of the underlying container. Another example is Facelets, a light-weight JSP replacment. I use JSF+Facelets on the OpenSSO project, and it has been flawless (the Facelets part). Why? Because I include the Facelets JAR, it is not provided by the system.
What if Java EE had followed the same model? Java EE should have been nothing more than a plugin framework and some base services. Want to use JSF in your project? Just include it from a managed, networked Java EE module repository. Include the one, common implementation of JSF. Include the approved version for your level of Java EE. If you support Java EE 5, you get JSF version X, across all web containers.