Re: EvalPlanQual behaves oddly for FDW queries involving system columns

On 2015/04/13 23:25, Jim Nasby wrote:
> On 4/13/15 4:58 AM, Etsuro Fujita wrote:
>> On 2015/04/10 21:40, Etsuro Fujita wrote:
>>> On 2015/04/09 12:07, Etsuro Fujita wrote:
>>>> I'll update the patch, which will contain only an infrastructure for
>>>> this in the PG core, and will not contain any postgres_fdw change.
>>>
>>> I updated the patch based on your comments. Updated patch attached. In
>>> the patch the following FDW APIs have been proposed:
>>>
>>> + RowMarkType
>>> + GetForeignRowMarkType (LockClauseStrength strength);
>>>
>>> + bool
>>> + LockForeignRow (EState *estate,
>>> + ExecRowMark *erm,
>>> + ItemPointer tupleid);
>>>
>>> + HeapTuple
>>> + FetchForeignRow (EState *estate,
>>> + ExecRowMark *erm,
>>> + ItemPointer tupleid);
>>>
>>> I think that these APIs allow the FDW that has TIDs to use the rowmark
>>> options such as ROW_MARK_REFERENCE, ROW_MARK_SHARE and
>>> ROW_MARK_EXCLUSIVE for its foreign tables so as to match the local
>>> semantics exactly, for example.
>>>
>>> As you mentioned, it would be better to add helper functions to see
>>> whether the foreign table is referenced by any ExecRowMarks. ISTM that
>>> an easy way to do that is to modify ExecFindRowMark() so that it allows
>>> for the missing case. I didn't contain such functions in the patch, though.
>>
>> I added that function and modified docs a bit. Please find attached an
>> updated version of the patch.
>
> Why aren't we allowing SELECT FOR KEY SHARE?

I omitted that mode (and the FOR NO KEY UPDATE mode) for simplicity, but
both modes have been allowed. However, I'm not sure if those modes are
useful because we don't have information about keys for a remote table.

Best regards,
Etsuro Fujita