[LLVMdev] Bringing back Andersen-like alias analysis

[Chris Lattner]

On Feb 19, 2012, at 4:18 PM, Tristan Schmelcher wrote:

> Hello,
> 
> I am researching a more efficient approach to subset-based alias analysis in the style of Andersen's algorithm, and I noticed that there used to be an implementation of that for LLVM back in 2.6 (-anders-aa) which was removed because it was "not being actively maintained and had substantial problems". I'd be interested in knowing what was wrong with it (other than Andersen's algorithm being O(n^3)) so that I can avoid the same mistakes. Does anyone recall what the issues were?

The major problems were that it was unmaintained and had a large number of bugs.  People kept trying to use it and were disappointed that it didn't work.  You can search bugzilla for closed bugs to see the ones that were open when andersens was removed.

-Chris