Re: SQL_DATA_AT_EXEC processing broken for large objects in nightly git

Hi Heikki,

The attached patch seems to fix all but I don't understand
sqltype_to_bind_pgtype() well.
Comments?

regards,
Hiroshi Inoue

On 2015/11/11 20:24, Inoue, Hiroshi wrote:
> On 2015/11/09 20:53, Heikki Linnakangas wrote:
>> On 11/04/2015 04:26 PM, Inoue, Hiroshi wrote:
>>>
>>>
>>> On 2015/11/04 20:04, Heikki Linnakangas wrote:
>>>> On 11/04/2015 09:48 AM, Inoue, Hiroshi wrote:
>>>>> Hi,
>>>>> To
>>>>> I was able to reproduce it here.
>>>>> Because I can't understand the current code at once, please give
>>>>> me some
>>>>> time
>>>>> to solve the problem.
>>>>
>>>> I also started looking at this last week, and came up with a patch
>>>> that I think fixes it, but I got distracted before completing it and
>>>> replying here. So here's what I came up with...
>>>
>>> The solution seems to generate 2 transactions for 1 insertion.
>>
>> Hmm, you're right, although that doesn't seem to be a new issue. The
>> 09.03.0400 driver does the same, see attached logs from
>> log_statements=all. Would be nice to fix that, but it's a separate
>> issue.
>
> Oh you are right. I will examine it.
>
>>
>>
>>> The insertion seems to be rejected when ByteaAsLongVarBinary=1.
>>
>> Oh. It works for me. How are you testing that?
>
> I define LO type and set ByteaAsLongVarBinary=1.
> The failure is #ERROR: column "large_data" is of type LO but
> expression is type "bytea" .
>
> regards,
> Hiroshi Inoue
>
>
>
>