While working on #11731 @revans2 noted that we're not 100% consistent on the use of null intolerance on GPU expressions. This probably doesn't hurt in practice since it's unlikely null intolerance will be considered after GPU plan replacements. However it would be nice to be consistent with the handling of null intolerance across all GPU expressions.
While working on #11731 @revans2 noted that we're not 100% consistent on the use of null intolerance on GPU expressions. This probably doesn't hurt in practice since it's unlikely null intolerance will be considered after GPU plan replacements. However it would be nice to be consistent with the handling of null intolerance across all GPU expressions.