Aha, that makes sense. Thanks for the pointer.
- Daniel p.s. Looks like the LLVM test suite shows no regressions from flipping the evaluator define. ----- Original Message ---- From: Eli Friedman <[EMAIL PROTECTED]> To: Daniel Dunbar <[EMAIL PROTECTED]> Cc: [email protected] Sent: Saturday, August 9, 2008 7:51:00 PM Subject: Re: [cfe-commits] r54598 - /cfe/trunk/lib/AST/Expr.cpp On Sat, Aug 9, 2008 at 7:28 PM, Daniel Dunbar <[EMAIL PROTECTED]> wrote: > Ok, this is better than what was before, but we still need to handle this case > if we want gems like > -- > enum x { v = (char*) 0 - (char*) 0 }; > -- > to behave as they do in GCC. I'm well aware of that issue; see http://lists.cs.uiuc.edu/pipermail/cfe-dev/2008-July/002232.html. Essentially, my proposal is to push the handling for this case outside of isICE. I haven't gotten around to implementing it yet, though. -Eli _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
