On Fri, 1 Dec 2023 at 14:30, David Rowley wrote:
>
> On Fri, 1 Dec 2023 at 13:14, Andres Freund wrote:
> > So I think adding an assert to ExecCopySlot(), perhaps with a comment saying
> > that the restriction could be lifted with a bit of work, would be fine.
>
> How about the attached? I wrote
On Fri, 1 Dec 2023 at 13:14, Andres Freund wrote:
> So I think adding an assert to ExecCopySlot(), perhaps with a comment saying
> that the restriction could be lifted with a bit of work, would be fine.
Thanks for looking at this again.
How about the attached? I wrote the comment you mentioned
Hi,
On 2023-11-06 11:16:26 +1300, David Rowley wrote:
> On Sat, 4 Nov 2023 at 15:15, Andres Freund wrote:
> >
> > On 2023-11-01 11:35:50 +1300, David Rowley wrote:
> > > I changed the Assert in tts_virtual_copyslot() to check the natts
> > > match in each of the slots and all of the regression te
On Sat, 4 Nov 2023 at 15:15, Andres Freund wrote:
>
> On 2023-11-01 11:35:50 +1300, David Rowley wrote:
> > I changed the Assert in tts_virtual_copyslot() to check the natts
> > match in each of the slots and all of the regression tests still pass,
> > so it seems we have no tests where there's an
Hi,
On 2023-11-01 11:35:50 +1300, David Rowley wrote:
> Looking at the Assert inside tts_virtual_copyslot(), it does:
>
> Assert(srcdesc->natts <= dstslot->tts_tupleDescriptor->natts);
I think that assert was intended to be the other way round.
> So, that seems to indicate that it's ok for the
Looking at the Assert inside tts_virtual_copyslot(), it does:
Assert(srcdesc->natts <= dstslot->tts_tupleDescriptor->natts);
So, that seems to indicate that it's ok for the src slot to have fewer
attributes than the destination. The code then calls
tts_virtual_clear(dstslot), then slot_getallatt
