Re: New 9.5.0100 does not work properly - Mailing list pgsql-odbc

From Walter Willmertinger
Subject Re: New 9.5.0100 does not work properly
Date
Msg-id CAHbMG0U+dVBXtjqf6qkffhx4gR4sop780sBc6dntpcDeiaTkeA@mail.gmail.com
Whole thread Raw
In response to Re: New 9.5.0100 does not work properly  ("Inoue, Hiroshi" <h-inoue@dream.email.ne.jp>)
Responses Re: New 9.5.0100 does not work properly  ("Inoue, Hiroshi" <h-inoue@dream.email.ne.jp>)
List pgsql-odbc
I switched on "Server side prepare" and it is working better.
But in our VBA we programmed "Boolean" as "Char". So we switched this to "On" in settings.
Now we have no write conflict, but we get this message:

Inline image 1

For testing purpose, I switched "Bool as Char" to "Off" and it seems to work with "Server side prepare".

So we will switch back to 9.3, as 9.4 and 9.5 are not working!

Regards
Walter Willmertinger


Inoue, Hiroshi <h-inoue@dream.email.ne.jp> schrieb am Mo., 18. Jan. 2016 um 13:38 Uhr:
Hi Walter and Heikki,

Probably I found the cause.


On 2016/01/12 17:36, Walter Willmertinger wrote:
The "Write Conflict" is independent of the setting "row versioning". When I strted it was switched off. I switched to "On", but the same result.

Here is an excerpt from mylog:

[7.065]conn=000001F55B153090, query='SELECT "KundenCode","DatumEreignis","AufnahmeDurchPersonalNr","WeiterAnPersonalNr","Status","Notiz","KdID","Prioritaet","ProgErledigt","update_cs","ZuletztBearbeitetPersonalNr" FROM "public"."Notizen" WHERE "KundenCode" = E'1' AND "KdID" = 83'

Doesn't the target table contain boolean items?
Then the driver has to handle parse and describe message before executing the
update query.


[8.455]ParseAndDescribeWithLibpq: plan_name= query=UPDATE "public"."Notizen" SET "Notiz"=$1 WHERE "KundenCode" = $2 AND "KdID" = $3 AND "DatumEreignis" = $4 AND "AufnahmeDurchPersonalNr" = $5 AND "WeiterAnPersonalNr" = $6 AND "Status" = $7 AND "Prioritaet" IS NULL AND "ProgErledigt" = $8 AND "update_cs" = $9 AND "ZuletztBearbeitetPersonalNr" IS NULL
[8.456]conn=000001F55B153090, query='BEGIN'
[8.457]ParseWithLibpq: plan_name= query=UPDATE "public"."Notizen" SET "Notiz"=$1 WHERE "KundenCode" = $2 AND "KdID" = $3 AND "DatumEreignis" = $4 AND "AufnahmeDurchPersonalNr" = $5 AND "WeiterAnPersonalNr" = $6 AND "Status" = $7 AND "Prioritaet" IS NULL AND "ProgErledigt" = $8 AND "update_cs" = $9 AND "ZuletztBearbeitetPersonalNr" IS NULL

But a simple query is issued without using extended protocol because the
*Server side prepare* option is turned off.


[8.460]conn=000001F55B153090, query='UPDATE "public"."Notizen" SET "Notiz"=E'Test xxx' WHERE "KundenCode" = E'1' AND "KdID" = 83 AND "DatumEreignis" = E'2015-09-30 13:48:13'::timestamp AND "AufnahmeDurchPersonalNr" = 80 AND "WeiterAnPersonalNr" = 80 AND "Status" = E'E' AND "Prioritaet" IS NULL AND "ProgErledigt" = E'0' AND "update_cs" = E'2015-09-30 13:48:15'::timestamp AND "ZuletztBearbeitetPersonalNr" IS NULL'

Though this update returns the response 'UPDATE 1', SQLRowCount() doesn't
work as expected due to SC_set_Result() in prepareParameters().

[8.464]conn=000001F55B153090, query='ROLLBACK'


One way to avoid the bug is to turn on the *Server side prepare*
option.
Though an appropriate bug fix is not clear to me, the attached
patch is an example.

regards,
Hiroshi Inoue
Attachment

pgsql-odbc by date:

Previous
From: Alvaro Herrera
Date:
Subject: [Raiford@labware.com: Re: float8 column size defined as 15 instead of 53?]
Next
From: "Inoue, Hiroshi"
Date:
Subject: Re: Re: 9.0.5 issue - duplicate rows returned when using SQL_ATTR_ROW_ARRAY_SIZE attribute