Thread

1. Re: [HACKERS] 6.5 cvs: views doesn't survives after pg_dump (fwd)

   Keith Parks <emkxp01@mtcc.demon.co.uk> — 1999-05-24T13:21:07Z

   Jan,
   
   Have you any ideas on this?
   
   We get a rule output by pg_dump like :-
   
   CREATE RULE "_RETsongs" AS 
     ON SELECT TO "songs" 
     WHERE  DO INSTEAD 
       SELECT "t"."artist", "t"."song", "t"."trackno", "d"."cdname" 
       FROM "disks" "d", "tracks" "t" 
       WHERE "d"."diskid" = "t"."diskid"; 
   
   from a view defined like so:-
   
   CREATE VIEW songs AS
     SELECT t.artist, t.song, t.trackno, d.cdname
     FROM disks d, tracks t
     WHERE d.diskid = t.diskid;
   
   Note the WHERE keyword in line 3 of the rule define.
   
   >From "./src/backend/utils/adt/ruleutils.c" line 662 of 1814
   
           /* If the rule has an event qualification, add it */
           if (ev_qual == NULL)
                   ev_qual = "";
           if (strlen(ev_qual) > 0)
           {
                   Node       *qual;
                   Query      *query;
                   QryHier    qh;                    
   .
   .
                   strcat(buf, " WHERE ");
                   strcat(buf, get_rule_expr(&qh, 0, qual, TRUE));
           }
   
           strcat(buf, " DO ");
   
           /* The INSTEAD keyword (if so) */
           if (is_instead)
                   strcat(buf, "INSTEAD ");        
   
   We put the WHERE in if strlen(ev_qual) > 0
   
   I've not yet followed this back any further.
   
   Keith.
    
   
   ------------ Begin Forwarded Message -------------
   
   X-Authentication-Warning: hub.org: majordom set sender to 
   owner-pgsql-hackers@postgreSQL.org using -f
   Date: Fri, 21 May 1999 22:34:50 +0100 (BST)
   From: Keith Parks <emkxp01@mtcc.demon.co.uk>
   Subject: Re: [HACKERS] 6.5 cvs: views doesn't survives after pg_dump (fwd)
   To: pgsql-hackers@postgreSQL.org, oleg@sai.msu.su
   MIME-Version: 1.0
   Content-MD5: 34XqWKKsmVlyonlE1gsMzw==
   
   Oleg Bartunov < oleg@sai.msu.su>
   > After dumping (by pg_dump) and restoring views becomes a tables
   > 
   
   The problem is that views are dumped with anm extraneous "WHERE"
   
   > ............................
   > QUERY: COPY "t1" FROM stdin;
   > CREATE RULE "_RETv1" AS ON SELECT TO "v1" WHERE  DO INSTEAD SELECT "a" FROM 
   "t1";
   > QUERY: CREATE RULE "_RETv1" AS ON SELECT TO "v1" WHERE  DO INSTEAD SELECT "a" 
   FROM "t1";
   
   ...................................................++++++
   
   > ERROR:  parser: parse error at or near "do"
   > EOF
   
   Which causes this error and the rule (View) is not Created.
   
   I don't know how the where clause gets in there but if you
   edit the dump before restoring all is OK.
   
   Keith.
   
   
   
   ------------- End Forwarded Message -------------
   
   
   
   

2. Re: [HACKERS] 6.5 cvs: views doesn't survives after pg_dump (fwd)

   Jan Wieck <jwieck@debis.com> — 1999-05-25T08:14:22Z

   >
   > Jan,
   >
   > Have you any ideas on this?
   
       Yepp
   
   >
   > We get a rule output by pg_dump like :-
   >
   > CREATE RULE "_RETsongs" AS
   >   ON SELECT TO "songs"
   >   WHERE  DO INSTEAD
   >     SELECT "t"."artist", "t"."song", "t"."trackno", "d"."cdname"
   >     FROM "disks" "d", "tracks" "t"
   >     WHERE "d"."diskid" = "t"."diskid";
   >
   > from a view defined like so:-
   >
   > CREATE VIEW songs AS
   >   SELECT t.artist, t.song, t.trackno, d.cdname
   >   FROM disks d, tracks t
   >   WHERE d.diskid = t.diskid;
   >
   > Note the WHERE keyword in line 3 of the rule define.
   >
   > >From "./src/backend/utils/adt/ruleutils.c" line 662 of 1814
   >
   >         /* If the rule has an event qualification, add it */
   >         if (ev_qual == NULL)
   >                 ev_qual = "";
   >         if (strlen(ev_qual) > 0)
   >         {
   >                 Node       *qual;
   >                 Query      *query;
   >                 QryHier    qh;
   > .
   > .
   
       That's   exactly   the   location  AFAICS.  The  problem  was
       introduced when the storage of  rules  changed  in  that  the
       event  qualification is now stored as "<>" (the output of the
       node print functions for NULL) instead of a NULL attribute.
   
       I'll fix it soon - thanks.
   
   >
   > We put the WHERE in if strlen(ev_qual) > 0
   >
   > I've not yet followed this back any further.
   >
   > Keith.
   
   
   Jan
   
   --
   
   #======================================================================#
   # It's easier to get forgiveness for being wrong than for being right. #
   # Let's break this rule - forgive me.                                  #
   #======================================== jwieck@debis.com (Jan Wieck) #