My Projects
taevals
taevals is a Google AppEngine application I wrote when I was the Lead TA to lessen the burden of completing the midterm TA evaluations. An important factor of TA evaluations is anonymity and thus taevals was designed with anonymity in mind. While it is easy to say an application should be trusted, I took this one step further and made the code available on Google Code under the New BSD License. More information about taevals can be found on its Google Code page.
turnin_helper
turnin_helper is an incredibly useful tool I wrote to assist in grading course projects. It is intended to work in tandem with UCSB CS's turnin program. See the README in the tarball for everything you'll need to know.
Features
- Automatic extraction of students' latest submission
- Execute make (of any target) on extracted submissions
- Email all students a private message about their project appended with a generic message about the project
Updates
- 2009-10-23 -- Added support for hyphens ("-") in user names. Verified against all user names currently in /cs/student and /cs/guest. turnin_helper will throw a warning if the user name has a hyphen followed by a number thus requiring manual extraction and or building of that particular user's submission.
Source
Download turnin_helper (tar.gz)
Optionally if you are a CS graduate student at UCSB you can clone the
mercurial source via:
hg clone ssh://<user>@csil.cs.ucsb.edu/~bboe/seclab/repos/turnin_helper
auto_grade
I wrote auto_grade in Winter 2009 following writing turnin_helper to minimize my effort in grading student's projects for CS160, Compilers. Though auto_grade reduces grading time a great deal of effort is needed at the start of a project to define the inputs and outputs. The primary benefit of auto_grade is that it provides immediate feedback to the students in a manor that is completely objective.
Features
- Immediate extraction, build, and score of student's submission
- Scoring based on diff output and exit status codes
- Immediate feedback to students via email which can contain information such as number of test cases passed as well as a variable amount of diff lines
- Option to set max running time of program
- Ability to assign different point values for different inputs, thus supporting extra credit inputs
Download auto_grade (tar.gz)
CRAWL-E
CRAWL-E is a web crawling framework that seamlessly supports distributed crawling across multiple threads as well as multiple machines. I wrote CRAWL-E in order to collect data from both Overstock.com and Facebook when I was working in Professor Ben Zhao's lab. Since that time, I have continued to improve CRAWL-E as issues appear and features are requested. CRAWL-E is released under the open source New BSD License. More information about CRAWL-E can be found on its Google Code page.
