Thread: remove redundant check of item pointer
In function ItemPointerEquals, the ItemPointerGetBlockNumber already checked the ItemPointer if valid, there is no need to check it again in ItemPointerGetOffset, so use ItemPointerGetOffsetNumberNoCheck instead. -- Regards Junwang Zhao
Attachment
Junwang Zhao <zhjwpku@gmail.com> writes: > In function ItemPointerEquals, the ItemPointerGetBlockNumber > already checked the ItemPointer if valid, there is no need > to check it again in ItemPointerGetOffset, so use > ItemPointerGetOffsetNumberNoCheck instead. I do not think this change is worth making. The point of ItemPointerGetOffsetNumberNoCheck is not to save some cycles, it's to be able to fetch the offset field in cases where it might validly be zero. The assertion will be compiled out anyway in production builds --- and even in assert-enabled builds, I'd kind of expect the compiler to optimize away the duplicated tests. regards, tom lane
got it, thanks for the explanation. On Wed, Apr 27, 2022 at 11:34 PM Tom Lane <tgl@sss.pgh.pa.us> wrote: > > Junwang Zhao <zhjwpku@gmail.com> writes: > > In function ItemPointerEquals, the ItemPointerGetBlockNumber > > already checked the ItemPointer if valid, there is no need > > to check it again in ItemPointerGetOffset, so use > > ItemPointerGetOffsetNumberNoCheck instead. > > I do not think this change is worth making. The point of > ItemPointerGetOffsetNumberNoCheck is not to save some cycles, > it's to be able to fetch the offset field in cases where it might > validly be zero. The assertion will be compiled out anyway in > production builds --- and even in assert-enabled builds, I'd kind > of expect the compiler to optimize away the duplicated tests. > > regards, tom lane -- Regards Junwang Zhao