[LLVMdev] RFC: Proposal for Poison Semantics

[Sanjoy Das]

Hi Bruce,

I think I misread your email.  Basically I wanted to say that you
cannot make a semantic difference between "and X, 0" and "and X, Y"
with Y's dynamic value being zero -- they need to have the exact same
semantics w.r.t. poison.  But on re-reading your mail, it does not
look like you were suggesting that.

Sorry for the confusion.
-- Sanjoy