llvm dev - May 2019 - Different internal representation of SwitchInst

SwitchInst is currently represented as paired operands, which is
internally a linked-list. This is not ideal, as these types are fixed
(APInt and BasicBlock), and there is also code of dubious quality that
doesn't really get that this is a linked list, like from
SwitchInst:removeCase:

  OL[NumOps-2].set(nullptr);
  OL[NumOps-2+1].set(nullptr);

I've already done quite a bit of work on the SwitchInst, and have much
more planned. I think the cases would be better represented by a
SmallArray<std::pair<APInt, BasicBlock*>, 4>.

Before I get time to jump in implement this, assuming I can probably
keep most of the interfaces, is there any reason this wouldn't work?

-Shawn Landden

This sounds like a great improvement to me Shawn!

-Chris
> On May 23, 2019, at 7:29 AM, Shawn Landden via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
> 
> SwitchInst is currently represented as paired operands, which is
> internally a linked-list. This is not ideal, as these types are fixed
> (APInt and BasicBlock), and there is also code of dubious quality that
> doesn't really get that this is a linked list, like from
> SwitchInst:removeCase:
> 
>  OL[NumOps-2].set(nullptr);
>  OL[NumOps-2+1].set(nullptr);
> 
> I've already done quite a bit of work on the SwitchInst, and have much
> more planned. I think the cases would be better represented by a
> SmallArray<std::pair<APInt, BasicBlock*>, 4>.
> 
> Before I get time to jump in implement this, assuming I can probably
> keep most of the interfaces, is there any reason this wouldn't work?
> 
> -Shawn Landden
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev