Re: Can rs_cindex be < 0 for bitmap heap scans?
От | Dilip Kumar |
---|---|
Тема | Re: Can rs_cindex be < 0 for bitmap heap scans? |
Дата | |
Msg-id | CAFiTN-spF4eWDZ1w=5EQ7E+S+JXtqB4FYGF9ECUFo6Cd51QV4Q@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Can rs_cindex be < 0 for bitmap heap scans? (Dilip Kumar <dilipbalaut@gmail.com>) |
Ответы |
Re: Can rs_cindex be < 0 for bitmap heap scans?
|
Список | pgsql-hackers |
On Fri, Oct 25, 2024 at 10:07 PM Melanie Plageman <melanieplageman@gmail.com> wrote:
On Fri, Oct 25, 2024 at 10:29 AM Melanie Plageman
<melanieplageman@gmail.com> wrote:
>
> Tom suggested off-list that if rs_cindex can't be zero, then it should
> be unsigned. I checked the other scan types using the
> HeapScanDescData, and it seems none of them use values of rs_cindex or
> rs_ntuples < 0. As such, here is a patch making both rs_ntuples and
> rs_cindex unsigned.
{
HeapTuple tuple = &(scan->rs_ctup);
Page page;
- int lineindex;
- int linesleft;
+ uint32 lineindex;
+ uint32 linesleft;
IMHO we can't make "lineindex" as uint32, because just see the first code block[1] of heapgettup_pagemode(), we use this index as +ve (Forward scan )as well as -ve (Backward scan).
[1]
if (likely(scan->rs_inited))
{
/* continue from previously returned page/tuple */
page = BufferGetPage(scan->rs_cbuf);
lineindex = scan->rs_cindex + dir;
if (ScanDirectionIsForward(dir))
{
/* continue from previously returned page/tuple */
page = BufferGetPage(scan->rs_cbuf);
lineindex = scan->rs_cindex + dir;
if (ScanDirectionIsForward(dir))
--Refer definition of ScanDirection
typedef enum ScanDirection
{
BackwardScanDirection = -1,
NoMovementScanDirection = 0,
ForwardScanDirection = 1
} ScanDirection;
{
BackwardScanDirection = -1,
NoMovementScanDirection = 0,
ForwardScanDirection = 1
} ScanDirection;
В списке pgsql-hackers по дате отправления: