martinboehme wrote: Update: Here's a draft implementation of the "simpler, bolder approach" referenced above:
https://github.com/llvm/llvm-project/pull/72985 The code is a _lot_ simpler -- it's mainly deleting code -- and I think I'm convinced at this point that this is the preferable thing to do. Would appreciate input. If others agree, there's not a whole lot of work remaining on this draft -- main thing is to see how this affects SAT solver timeouts on real-world codebases (I expect the effect to be similar to this current patch) and some simplification in the code (converting two function templates to non-template functions because they now only get called with one specific set of argument types). https://github.com/llvm/llvm-project/pull/72850 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
