Thread

1. Maybe a plpgsql bug?

   Lehel Gyuro <lehel@bin.hu> — 2001-04-10T07:31:32Z

   I've tried to write a plpgsql function, and noticed the following
   problem : (7.1RC2 rpm from postgresql.org)
   
   When issuing a comand like:
   DECLARE
     rowvar tablaname%ROWTYPE
   BEGIN
     FOR rowvar IN SELECT FROM tablename t WHERE t.xxx=yyy AND t.zzz=qqq
     LOOP
     .....
     END LOOP;
   
   or just simply issuing a query like
   
   DECLARE
     rownum int4:=0;
   BEGIN
     SELECT count (*)::int4 INTO rownum FROM tablename t WHERE t.xxx=yyy
     AND t.zzz=qqq;
   
   only the t.zzz=qqq is realized in the returned rows, the t.xxx=yyy is
   skipped, but when executing the query with execute everything is fine.
   Is this an undocumented bug? I've noticed that the same query
   structure is avoided in the documentation, you use simple queryes with
   only one filtering condition... :(
   
   
   
   

2. Re: Maybe a plpgsql bug?

   Tom Lane <tgl@sss.pgh.pa.us> — 2001-04-13T17:22:38Z

   Lehel Gyuro <lehel@bin.hu> writes:
   > I've tried to write a plpgsql function, and noticed the following
   > problem : (7.1RC2 rpm from postgresql.org)
   
   Perhaps this is a bug, but you have not given a complete example that
   would allow someone else to try to reproduce it.  Please see the
   guidelines for bug reports.
   
   			regards, tom lane