steakhal wrote:

> What kind of effort is needed to make the z3 backend suitable to be supported?

I think, at minimum, we should be able to set Z3 as the default solver and run 
all the tests we have in `check-clang-analysis`. It will fail them of course 
because it doesn't meet the expectations but it must not crash.

Once we have that, we can look at the open GH issues and check if any of those 
are still valid and fix them.
After that, we should try to analyze clang itself, and if it still crashes, fix 
those too.

After all this, I think we would have someone who built up enough experience in 
that part of the codebase to say that we have expertise. The last remaining 
component would be have someone around who is willing to invest into triaging 
and potentially fixing incoming issues wrt. the Z3 solver.

As a bonus, one could look into performance anomalies that the Z3 solver is 
known for - but I'd say that fixing crashes is more important than slow 
analysis or hangs.

https://github.com/llvm/llvm-project/pull/205370
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to