It’s really been two weeks since I’ve posted here. It’s also been two weeks since I’ve worked on my concurrent unit testing project. Scary.
Of course, it’s not that I’ve done nothing, far from it. The first week Corky and I started working on adding unit tests for the Javadoc component, but then got sidetracked by a request from University of Washington. They requested that the current directory (the “working directory”, basically new File(".")
) should be configurable, something I’ve requested numerous times before, particularly in connection with the Rice MBS.
On Wednesday of that week, we got the initial sketch for that change done, at least outside of project mode. For projects, the working directory should be configurable on a per-project basis and stored with the project file. Doing that took a few more days. Corky did most of the work.
When Corky submitted his changes at the beginning of the second week, I noticed that they were failing on my Linux configuration, but only there. I looked at what was happening, and finally on Wednesday realized that the changes we made don’t work if the working directory was set to a location that didn’t exist anymore. I made the necessary changes.
On Thursday, I wanted to write a unit test for this condition (because technically, this was a bug, so there should be a unit test exhibiting it), but found that the majority of the unit tests for DrJava were set up incorrectly. The setUp()
and tearDown()
methods didn’t call super.setUp()
and super.tearDown()
, so there was no one place where I could put a line of code that should be executed for all of the tests. With a lot of busy hand work, I changed that and did a massive commit on Friday.
I guess now I’ve been promoted to an “experienced person” in the DrJava team, so Corky and I won’t be working as a pair anymore, now Tim and I will handle small projects together. We’ve started to look at adding a “Go to file” dialog that does some sort of prediction based on what the user inputs.