Обсуждение: Question about building an exportable snapshop

While working on the issue [1], I realize that if a subtransaction
hasn't done any catalog change then we don't add this in the commit
xid list even if we are building a full snapshot [2].  That means when
we will convert this to the MVCC snapshot we will add this to a
running xid list.  If my understanding is correct then how visibility
will work? Because if I look at the code in XidInMVCCSnapshot(), then
if the suboverflowed is not set then first we are going to look into
the snapshot->subxip array and if we don't find it there then we look
into the snapshot->xip array,  and now we will be finding even
committed subxips in the snapshot->xip array.  Am I missing something?

[2]
/*
* Add subtransaction to base snapshot if catalog modifying, we don't
* distinguish to toplevel transactions there.
*/
if (ReorderBufferXidHasCatalogChanges(builder->reorder, subxid))
{
sub_needs_timetravel = true;
needs_snapshot = true;

[3]

[1] https://www.postgresql.org/message-id/CAFiTN-tqopqpfS6HHug2nnOGieJJ_nm-Nvy0WBZ%3DZpo-LqtSJA%40mail.gmail.com

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

On Tue, Oct 12, 2021 at 6:21 PM Dilip Kumar <dilipbalaut@gmail.com> wrote:
>
> While working on the issue [1], I realize that if a subtransaction
> hasn't done any catalog change then we don't add this in the commit
> xid list even if we are building a full snapshot [2].
>

I think this is true only if we have reached SNAPBUILD_CONSISTENT
state otherwise, we are adding subtransactions in the committed xip
array by setting 'needs_timetravel' to true. And if we have already
reached a consistent state before it then we anyway don't need to add
this. If this is true, do you still see any problem?

-- 
With Regards,
Amit Kapila.

On Wed, Oct 20, 2021 at 5:06 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Tue, Oct 12, 2021 at 6:21 PM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> >
> > While working on the issue [1], I realize that if a subtransaction
> > hasn't done any catalog change then we don't add this in the commit
> > xid list even if we are building a full snapshot [2].
> >
>
> I think this is true only if we have reached SNAPBUILD_CONSISTENT
> state otherwise, we are adding subtransactions in the committed xip
> array by setting 'needs_timetravel' to true. And if we have already
> reached a consistent state before it then we anyway don't need to add
> this. If this is true, do you still see any problem?

Yeah, you are right.

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com