[plt-scheme] Whence MrSpidey?

On Jan 10, 2007, at 4:20 PM, Brent Fulgham wrote:

> Googling today for Lisp static analysis tools, I re-discovered the  
> MrSpidey stuff from older DrScheme implementations.  Was MrSpidey  
> jettisoned in favor of newer features, or was it simply an  
> evolutionary dead-end that was removed due to lack of interest?

MrSpidey was jettisoned due to two things:

1. The code wasn't maintained. Neither the original creator nor the  
then-research scientist with the project were able/interested/willing  
to maintain the code as the basis went thru significant changes.  
Worse, we would have had to change Spidey (as we lovingly called it)  
to cope with a new language. In a sense, the evolution of the  
language should have made it easier but it just didn't work out.

2. We discovered some basic algorithmic flaws that made Spidey far  
less accurate than we wanted it. (It signaled too many false  
positives.)

Taking 1 and 2 together, we started a new effort -- tuned to the  
modular version of PLT Scheme -- and got, well, not as far as we  
wanted. Philippe Meunier, the PhD student who tackled this problem,  
finished the theory (see POPL 2006) but not the practice. Since it  
was time for him to move on, we're stuck right now. He doesn't intend  
to expand his prototype (in svn) to apply to the full language one  
day, and I bet he'd love to have volunteer work but -- being a member  
of the French Foreign Academic Legion -- he is sitting in Bangkok on  
a tank grading exams instead of hacking on Analyze (the new name for  
the tool).

Applications welcome: meunier at ccs.neu.edu.

-- Matthias