Thoughts on agile software development
FreshMarker is an embedded Java template engine that focusses on extensibility and simplicity. For this reason, every implementation of an extension is associated with weighing up its pros and cons. For a long time, the import directive therefore remained an unrealised feature.
While reviewing the mutation tests for FreshMarker, I noticed an implementation gap in the logical expression. So if something needs to be changed, why not quickly add something new.
A Java developer always likes to look into the JDK, like a magician in his old tomes, to see if there is an unknown spell, er method, with which he can improve his source code. In fact, Java 21 has introduced a lot of new methods that make everyday life a little easier.
With Java 22, string interpolation is fighting for its place in the Java ecosystem. What works in other languages and many frameworks should now also simplify the work of Java developers. Some wishes are fulfilled for the developers, others unfortunately not and some would never have been expressed. But more on that later.
It is always instructive to look at well-known open source projects to learn from the implementation used. Sometimes you look at the unit tests and start to wonder. A serious problem with large and old frameworks is the time-consuming adaptation to changing environments. Modern features in libraries such as JUnit 5 simplify the code base immensely.
Some additions to a framework are so trivial that they don’t really deserve their own article. On the other hand, as a developer, you may ask yourself why this addition has taken so long. For unknown data types, FreshMarker uses the interpretation of the class as a Java Bean by default. This is a good solution for many POJO, DTO and Entity classes, but in some cases the class should be used as a primitive FreshMarker type.
The Rocinante library has now passed some important milestones. It can read and write messages and masters the scalar base types. The next step is now to support Repeated Fields. Repeated fields are the Protocol Buffer counterpart to lists and arrays in other protocols.
After the Rocinante project has mastered some basic functionalities of Protocol Buffer, the project automation is now to be improved by a Maven plugin.
Sometimes even developers want to present something. It’s nice when there’s a tool that picks up the developers in their comfort zone. Software developers are a special clientele and have very specific requirements for a software solution for creating and giving presentations.
The third article on Rocinante finally deals with the interesting part of the protocol. How the parts of a message are encoded into the binary Protocol Buffer representation and decoded again. Rocinante uses a two-stage approach. The basis is formed by a ProtoInputStream and a ProtoOutputStream, in which the basic mechanisms are implemented. For each message a generated Proto class with individual read and write methods is based on this.