Re: [ANNOUNCE] psqlODBC 09.03.0300 Released

On Tuesday 27 of May 2014 16:20:49 Heikki Linnakangas wrote:
> On 05/26/2014 02:11 AM, Michael Paquier wrote:
> > On Mon, May 26, 2014 at 6:17 AM, Christoph Berg <cb@df7cb.de> wrote:
> >> Re: Hiroshi Saito 2014-05-18 <537816AF.2070308@winpg.jp>
> >> ============== running regression test queries        ==============
> >> test catalogfunctions         ... FAILED
> >> The missing lines were introduced in
> >> ad2a4aecb4748f8c5280aabf65dae570a9529065. Possibly the generator for
> >> these wasn't updated/committed along?
> > This is being discussed here as well:
> > http://www.postgresql.org/message-id/CAB7nPqQN=CB2HXwB5DRAgHwhMnHQBHh2QaNPFbMB+z_ohsVY_w@mail.gmail.com
> > I think that this portion of the commit should be reverted. Saito-san
> > also did not provide the conditions (OS, configuration) under which he
> > can reproduce this error.
> >
> > Output as it stands now is wrong btw as SQLPrivileges should not
> > return the same privileges twice for the same table...
>
> Yeah, it's definitely wrong as it is. Since we haven't heard from
> Saito-san why he did that, I'm assuming it was an oversight. I just
> reverted the bogus change to the expected output, so it should work now.

FWIW, tested with PostgreSQL 9.3.5 and I see there the privileges twice,
there are not same though.  I tried this:

        diff --git a/src/catalogfunctions-test.c b/src/catalogfunctions-test.c
        index d2fbc34..2ff162d 100644
        --- a/src/catalogfunctions-test.c
        +++ b/src/catalogfunctions-test.c
        @@ -25,7 +25,7 @@ main(int argc, char **argv)
                int rc;
                HSTMT hstmt = SQL_NULL_HSTMT;
                /* Cases where output is limited to relevant information only */
        -       SQLSMALLINT sql_tab_privileges_ids[6] = {1, 2, 3, 4, 6, 7};
        +       SQLSMALLINT sql_tab_privileges_ids[7] = {1, 2, 3, 4, 5, 6, 7};
                SQLSMALLINT sql_column_ids[6] = {1, 2, 3, 4, 5, 6};

                test_connect();
        @@ -180,7 +180,7 @@ main(int argc, char **argv)
                                                                (SQLCHAR *) "testtab1", SQL_NTS);
                CHECK_STMT_RESULT(rc, "SQLTablePrivileges failed", hstmt);
                print_result_meta(hstmt);
        -       print_result_series(hstmt, sql_tab_privileges_ids, 6);
        +       print_result_series(hstmt, sql_tab_privileges_ids, 7);
                rc = SQLFreeStmt(hstmt, SQL_CLOSE);
                CHECK_STMT_RESULT(rc, "SQLFreeStmt failed", hstmt);

And I got this:

      *** /tmp/odbc-test/upstream/test/expected/catalogfunctions.out  2014-11-18 12:23:24.000000000 +0100
      --- /tmp/odbc-test/upstream/test/results/catalogfunctions.out   2014-11-18 15:18:03.530186354 +0100
      ***************
      *** 184,194 ****
        PRIVILEGE: VARCHAR(128) digits: 0, nullable
        IS_GRANTABLE: VARCHAR(128) digits: 0, nullable
        Result set:
      ! contrib_regression    public  testtab1        _SYSTEM INSERT  YES
      ! contrib_regression    public  testtab1        _SYSTEM SELECT  YES
      ! contrib_regression    public  testtab1        _SYSTEM UPDATE  YES
      ! contrib_regression    public  testtab1        _SYSTEM DELETE  YES
      ! contrib_regression    public  testtab1        _SYSTEM REFERENCES      YES
        Result set:
        contrib_regression    public  testtab1        TABLE
        contrib_regression    public  testtab_fk      TABLE
      --- 184,199 ----
        PRIVILEGE: VARCHAR(128) digits: 0, nullable
        IS_GRANTABLE: VARCHAR(128) digits: 0, nullable
        Result set:
      ! contrib_regression    public  testtab1        _SYSTEM postgres        INSERT  YES
      ! contrib_regression    public  testtab1        _SYSTEM postgres        SELECT  YES
      ! contrib_regression    public  testtab1        _SYSTEM postgres        UPDATE  YES
      ! contrib_regression    public  testtab1        _SYSTEM postgres        DELETE  YES
      ! contrib_regression    public  testtab1        _SYSTEM postgres        REFERENCES      YES
      ! contrib_regression    public  testtab1        _SYSTEM test    INSERT  YES
      ! contrib_regression    public  testtab1        _SYSTEM test    SELECT  YES
      ! contrib_regression    public  testtab1        _SYSTEM test    UPDATE  YES
      ! contrib_regression    public  testtab1        _SYSTEM test    DELETE  YES
      ! contrib_regression    public  testtab1        _SYSTEM test    REFERENCES      YES
        Result set:
        contrib_regression    public  testtab1        TABLE
        contrib_regression    public  testtab_fk      TABLE

      ======================================================================

Seems like the testsuite is _required_ to be run under 'postgres' user,
isn't it truth?

Pavel