Re: Fwd: [BUGS] Small bug in psqlodbc-09.01 prevents interoperability with LISTSERV - Mailing list pgsql-odbc

Hi.

Sorry, very late reaction...

We are seeing and considering carefully from posting to pgsql-bugs.
However, this answering and correspondence will be very difficult now.
Probably, much time is needed anyhow. I want to be referred to as TODO
for the time being...

We want to release the version which Fixed many bug of Current HEAD.
Please don't think that it is bad, I appreciate your perseverance. thanks.

Regards,
Hiroshi Saito

(2012/08/08 10:16), Craig Ringer wrote:
> ODBC issue report from -bugs
>
>
> -------- Original Message --------
> Subject:     [BUGS] Small bug in psqlodbc-09.01 prevents interoperability
> with LISTSERV
> Date:     Sat, 4 Aug 2012 15:43:58 +0000
> From:     Eric Thomas <ERIC@LSOFT.COM>
> To:     pgsql-bugs@postgresql.org <pgsql-bugs@postgresql.org>
>
>
>
> Because we have customers who think Excel is an RDBMS, our product
> LISTSERV religiously queries ODBC capabilities at startup and prints
> warnings as appropriate. A customer tried to use PostgreSQL on Windows
> and got the following (these are our messages not yours, I am pasting
> them to show all the version numbers):
>
>>  Driver manager version: 03.80.7601.0000
>
>>  ODBC driver: PSQLODBC35W.DLL (09.01.0100)
>
>>  DBMS: PostgreSQL (9.1.4)
>
>>  [FATAL] LIKE operator has no ESCAPE clause, errors will occur!
>
>>  [SEVERE] FOR UPDATE clause not supported, no locking will occur
>
> Due to the popularity of underscores in e-mail addresses, we require
> support for LIKE … ESCAPE and issue a fatal compatibility warning if the
> functionality is not present in the DBMS. A fatal warning means that we
> do not support the DBMS and will not accept incidents involving it. In
> short, the customer must choose another database, in practice MySQL,
> /the/ most problematic database we have ever had to support L
>
> But from what I understand, PostgreSQL does support both the LIKE …
> ESCAPE clause and the SELECT … FOR UPDATE clause. What’s more, the
> ‘{escape}’ sequence seems to be implemented by the ODBC driver, but the
> driver claims otherwise for some reason. Here is the relevant code from
> info.c:
>
> case SQL_LIKE_ESCAPE_CLAUSE: /* ODBC 2.0 */
>
> /*
>
> * is there a character that escapes '%' and '_' in a LIKE
>
> * clause? not as far as I can tell
>
> */
>
> p = "N";
>
> break;
>
> I assume that this was true many years ago and someone forgot to change
> it when implementing the ‘{escape}’ sequence J For the second issue, the
> code is in info30.c:
>
> case SQL_FORWARD_ONLY_CURSOR_ATTRIBUTES1:
>
> len = 4;
>
> value = SQL_CA1_NEXT; /* others aren't allowed in ODBC spec */
>
> break;
>
> Here I am puzzled as the Microsoft documentation says no such thing, at
> least not today, but even in an earlier version of ODBC it would have
> made no sense. Allowing only the SQL_CA1_NEXT flag would defeat the
> whole purpose of this information call – which database would /not/
> support SQL_FETCH_NEXT? J Anyway, LISTSERV is looking for the flag
> SQL_CA1_SELECT_FOR_UPDATE. Note that the same problem exists for other
> types of cursors, although LISTSERV does not use dynamic cursors and
> only queries the capabilities forward-only cursors.
>
> _This is not a cosmetic issue_. LISTSERV will not use any ODBC features
> reported as unsupported, so we cannot tell customers that “it complains
> at startup but in practice it works fine” – it doesn’t. E-mail addresses
> containing underscores or percent signs will cause unwanted side effects
> because LISTSERV thinks that using ‘{escape}’ would result in a syntax
> error.
>
> For reference, MySQL used to have the same two problems, and showed no
> interest in fixing them because “You should not use ODBC.” Well how
> exactly am I supposed to support SQL Server and DB2? Wait until
> Microsoft and IBM switch to the mysql_xxx() API? J But Oracle provided a
> solution for the LIKE escape in the ODBC driver. The driver still
> reports lack of support for the FOR UPDATE clause though, probably
> because it is only supported by some of their umpteen back-end engines
> so it varies from one table to another :D
>
> Eric
>
>
>
>