Main menu:

Site search



NS_OK-always analysis update

First note: I forgot to mention earlier that the results of these analyses are being posted in Bugzilla as bug #407444.

And as I wrote there, the analysis worked, finding about 10,000 methods that always return NS_OK out of 30,000 methods that return nsresult. The constraint system took about 6 seconds to solve using my basic Python solver–my worries about solving time were unfounded. It took about an hour to generate the constraint system, because of my horribly inefficient database query representation of the call graph.

Now the question becomes, what to do with these methods.


Comment from Robert O’Callahan
Time: January 17, 2008, 5:17 pm

Get your blog added to!

Comment from dmandelin
Time: January 17, 2008, 5:30 pm

I filed a bug to have the blog added to planet mozilla on Tuesday. Maybe I need to bug someone more.

Comment from Manu
Time: January 17, 2008, 6:55 pm

Dude, you’re blogging! The work sounds cool. Just don’t make my research obsolete with a 5 line Python constraint solver 🙂

Comment from dmandelin
Time: January 17, 2008, 7:06 pm

Hey, Manu. I was just wondering what your new email address is. But apparently you’re still using your Berkeley address? And how about if I don’t make your stuff obsolete with my 5-line solver in exchange for an awesome points-to analysis for C++? 🙂

Comment from Manu
Time: January 18, 2008, 11:47 am

Hey, yeah, the old email should forward even if/when that account is disabled. I assumed your 5-line solver would be an awesome points-to analysis for C++ (if you just pass in the right parameters); maybe you need 6 lines 🙂