A minor update related to Issue 42 on the Issue Tracker.
Essentially, I want to make everything output as they should in Scheme as opposed to how they do in Kawa/Java. For each case below, the first option is what we see now and the second is what we should see.
To start off, I have a small list of bug fixes / minor improvements before I get to the real meat of what I’ve got for the day:
trace still doesn’t work#t and false to #f to make input and output matchToday saw another round of bug fixes:
Since I moved to Kawa, I’ve been having an issue with maintaining the state between calls to the interpreter. Everything works fine when I load and run an entire file at once (for example when loading the options files or actually debugging my own code), but when I try to run individual lines of code to add custom bindings, it doesn’t work.
Two big changes for the project today. I’ve changed the IDE I’m using from Netbeans to Eclipse. Then, the back-end scheme system will be changing from SISC to Kawa.
From the beginning, I wanted Wombat to be configurable where it mattered, so I added a system that would be able to load and save options. This being a Scheme IDE, I wrote the options files in… what else? Scheme! Here’s the default options.cfg file included with the current distribution of Wombat (r90). The code to read these options is defined using Scheme macros written partially in a hyrbid of Java and Scheme and can (at least theoretically) even be set during runtime by calling the cfg procedure in the REPL.
One of the features that I really wanted to have in Wombat was the ability to have multiple documents in separate windows with the ability to drag them around both in their parent window or out of it, docking as necessary. Most modern IDEs have this feature and I figured that I should make Wombat as easy to use and like the others as I could. I looked around somewhat at different options for docking windows in Java and ended up settling on InfoNode Docking Windows, available under the GPL v3. It took some work to adapt to their API, but within a few hours I was able to release another build, r87 using the new system, available here.
The second (and it turned out final) iteration of Wombat was based on Java. With nearly 10 years of Java experience at that point, I could write Java code more quickly and efficiently than most other languages. In addition, it had the advantage of being portable, running on essentially every modern operating system with no changes to the deployed class files.
For three semesters now, I have been an AI (associate instructor, not the other kind) for the C211: Introduction to Computer Science class at Indiana University. C211 is (and has been for many years) taught using the Scheme programming language, a dialect of Lisp following a minimalist viewpoint which makes it ideal for education. Unfortunately, many of the IDEs most commonly used to edit Scheme, Lisp, and other languages based on S-expressions tend to be rather complex and require time to learn on their own, reducing the time dedicated to other topics in such an introductory class.