Running Code Tests on the Server

This week I worked on integrating the back-end of Looking Glass code tests for mentors into the community. This has been not as straight forward as I had hoped, but only because the development setup and production setup are different. I didn't plan that I would have to make this work in both the desktop and the server which is taking more time than I would like. Besides that it is working quite well. I've got the Looking Glass IDE uploading code tests and then the community saving the code test and queuing a job. This job will then run in a separate process to execute the code test. I'm trying to get most of the back-end in place and test it all before the weekend. I cannot yet run the code tests because I need to talk about what gets recorded when a code tests run with Caitlin.

I've also been supporting Shannon on her project. Her project this summer is the develop personalized tutorials based on a users perceived skill set. We've started trying to make it so we can automatically generate additional steps to make users learn more during the tutorial. So far this has been really hard. Because of the way we crawl the AST and then generate transactions, if want to modify the order of these transactions later (or even during) the entire tutorial becomes messed up. This makes sense because we need to be able to specify retarget criteria when we do this. The bottom line, the current system does not make this easy... and this is something we will want to do a lot in the future. I'm now wondering if we should instead generate pseudo transactions which can do all this retargeting for us so if want to change things to make the tutorial more adaptive... we can!