Re: [BUG?] strange behavior in ALTER TABLE ... RENAME TO on inherited columns

KaiGai Kohei <kaigai@kaigai.gr.jp> writes:
> (2009/12/30 10:38), Robert Haas wrote:
>> No longer applies.  Can you rebase?

> The attached patch is the rebased revision.

I'm not really impressed with this patch, because it will reject
perfectly legitimate multiple-inheritance cases (ie, cases where there's
more than one inheritance path from the same parent).  This works fine
at the moment:

regression=# create table p1(f1 int, f2 int);
CREATE TABLE
regression=# create table c1(f3 int) inherits (p1);
CREATE TABLE
regression=# create table c2(f4 int) inherits (p1);
CREATE TABLE
regression=# create table cc(f5 int) inherits (c1,c2);
NOTICE:  merging multiple inherited definitions of column "f1"
NOTICE:  merging multiple inherited definitions of column "f2"
CREATE TABLE
regression=# \d cc
      Table "public.cc"
 Column |  Type   | Modifiers 
--------+---------+-----------
 f1     | integer | 
 f2     | integer | 
 f3     | integer | 
 f4     | integer | 
 f5     | integer | 
Inherits: c1,
          c2

regression=# alter table p1 rename f2 to ff2;
ALTER TABLE
regression=# \d cc
      Table "public.cc"
 Column |  Type   | Modifiers 
--------+---------+-----------
 f1     | integer | 
 ff2    | integer | 
 f3     | integer | 
 f4     | integer | 
 f5     | integer | 
Inherits: c1,
          c2


I don't think that protecting against cases where things won't work
is an adequate reason for breaking cases that do work.

			regards, tom lane