MultithreadedTC is a framework that makes it easier to test concurrent abstractions (as opposed to JUnit, which doesn’t play well with threads). It was mentioned in the talk by Pugh, Goetz (of Java Concurrency in Practice fame) and Click (of lock-free hash table fame) on Testing Concurrent Software at JavaOne 2007.
I spent a few hours during the last week downloading and building various Java codebases and testing them with IKVM.NET.
So far, I’ve tried:
Most of the projects used Apache Ant to build. A few used Maven. Some projects had an Ant rule to download dependencies (nice touch). Other projects included their dependencies inside the source repository. Overall, building the projects using Ant inside Eclipse was a breeze.
I think all the projects used JUnit as is, with the exception of Derby, which has a rather complicated test setup (probably for good reasons).
I managed to reduce two failing test cases to Classpath bugs, namely bug #31814 (already fixed) and bug #31819. Jeroen Frijters also found and fixed various bugs.
I also played with the idea of setting up a Hudson instance to do all this automatically in the future. Take a look. I’ll contact the Classpath folks next week to here if they want to provide a permanent home for this setup.
I watched the Google TechTalk Java on Guice: Dependency Injection, the Java Way today. Very interesting stuff. Definitely worth watching before you dive into the rest of the Guice documentation.
So you just downloaded some Java goodie from the Internet, read the documenation and tried to run it. You get the following error: java.lang.NoClassDefFoundError: and
What probably happened is that you extracted the Java goodie to somewhere under C:\Documents and Settings on Windows and somewhere something doesn’t quite understand long filenames or some path isn’t quoted correctly.
I recently cooked up a little Python script that uses the excellent BeautifulSoup HTML/XML parser to scrape the images in the swell, period and wind charts for South Africa from MagicSeaweed.com. I used the time series feature that was added in Google Earth 4 to display the data in sequence. You can view the results here:
If anyone knows where one could find the raw data from which these images were made, I’d be very interested. Plotting the data as polygons would probably look much nicer than image overlays when zooming in.
If anyone is interested in the code that does the scraping, let me know. You’ll only have to change a few coordinates (obtained using Google Earth) and URLs to adapt it to fetch the reports for your location.
Looks like EditDNS.net went down recently, taking a few of my domains with it. I hope they come back, since they were only free provider I could find that supported SRV records.
Update: More on the outage: