Обсуждение: pgsql: Instead of supposing (wrongly, in the general case) that the

Log Message:
-----------
Instead of supposing (wrongly, in the general case) that the rowtype
of an inheritance child table is binary-compatible with the rowtype of
its parent, invent an expression node type that does the conversion
correctly.  Fixes the new bug exhibited by Kris Shannon as well as a
lot of old bugs that would only show up when using multiple inheritance
or after altering the parent table.

Modified Files:
--------------
    pgsql/src/backend/executor:
        execQual.c (r1.169 -> r1.170)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/executor/execQual.c.diff?r1=1.169&r2=1.170)
    pgsql/src/backend/nodes:
        copyfuncs.c (r1.293 -> r1.294)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/nodes/copyfuncs.c.diff?r1=1.293&r2=1.294)
        equalfuncs.c (r1.232 -> r1.233)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/nodes/equalfuncs.c.diff?r1=1.232&r2=1.233)
        outfuncs.c (r1.243 -> r1.244)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/nodes/outfuncs.c.diff?r1=1.243&r2=1.244)
        readfuncs.c (r1.173 -> r1.174)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/nodes/readfuncs.c.diff?r1=1.173&r2=1.174)
    pgsql/src/backend/optimizer/prep:
        prepjointree.c (r1.23 -> r1.24)

(http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/optimizer/prep/prepjointree.c.diff?r1=1.23&r2=1.24)
        prepunion.c (r1.117 -> r1.118)

(http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/optimizer/prep/prepunion.c.diff?r1=1.117&r2=1.118)
    pgsql/src/backend/optimizer/util:
        clauses.c (r1.184 -> r1.185)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/optimizer/util/clauses.c.diff?r1=1.184&r2=1.185)
    pgsql/src/backend/parser:
        parse_coerce.c (r2.124 -> r2.125)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/parser/parse_coerce.c.diff?r1=2.124&r2=2.125)
        parse_expr.c (r1.176 -> r1.177)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/parser/parse_expr.c.diff?r1=1.176&r2=1.177)
    pgsql/src/backend/utils/adt:
        ruleutils.c (r1.185 -> r1.186)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/backend/utils/adt/ruleutils.c.diff?r1=1.185&r2=1.186)
    pgsql/src/include/nodes:
        execnodes.h (r1.120 -> r1.121)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/include/nodes/execnodes.h.diff?r1=1.120&r2=1.121)
        nodes.h (r1.161 -> r1.162)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/include/nodes/nodes.h.diff?r1=1.161&r2=1.162)
        primnodes.h (r1.104 -> r1.105)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/include/nodes/primnodes.h.diff?r1=1.104&r2=1.105)
    pgsql/src/pl/plpgsql/src:
        pl_exec.c (r1.123 -> r1.124)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/src/pl/plpgsql/src/pl_exec.c.diff?r1=1.123&r2=1.124)

Tom Lane wrote:
> Instead of supposing (wrongly, in the general case) that the rowtype
> of an inheritance child table is binary-compatible with the rowtype of
> its parent, invent an expression node type that does the conversion
> correctly.  Fixes the new bug exhibited by Kris Shannon as well as a
> lot of old bugs that would only show up when using multiple inheritance
> or after altering the parent table.

Attached is a patch adding regression tests for this change. I'll apply
it to HEAD later tonight or tomorrow baring any objections. I just used
the two test cases posted in the -hackers thread; Tom, if you have any
other test cases for this behavior in mind, just let me know.

-Neil
--- src/test/regress/expected/inherit.out
+++ src/test/regress/expected/inherit.out
@@ -623,3 +623,31 @@
  32 | one | two | three
 (1 row)

+-- Tests for casting between the rowtypes of parent and child
+-- tables. See the pgsql-hackers thread beginning Dec. 4/04
+create table base (i integer);
+create table derived () inherits (base);
+insert into derived (i) values (0);
+select derived::base from derived;
+ derived
+---------
+ (0)
+(1 row)
+
+drop table derived;
+drop table base;
+create table p1(ff1 int);
+create table p2(f1 text);
+create function p2text(p2) returns text as 'select $1.f1' language sql;
+create table c1(f3 int) inherits(p1,p2);
+insert into c1 values(123456789, 'hi', 42);
+select p2text(c1.*) from c1;
+ p2text
+--------
+ hi
+(1 row)
+
+drop function p2text(p2);
+drop table c1;
+drop table p2;
+drop table p1;
--- src/test/regress/sql/inherit.sql
+++ src/test/regress/sql/inherit.sql
@@ -146,7 +146,25 @@

 -- Test changing the type of inherited columns
 insert into d values('test','one','two','three');
+alter table a alter column aa type integer using bit_length(aa);
+select * from d;

-alter table a alter column aa type integer using bit_length(aa);
+-- Tests for casting between the rowtypes of parent and child
+-- tables. See the pgsql-hackers thread beginning Dec. 4/04
+create table base (i integer);
+create table derived () inherits (base);
+insert into derived (i) values (0);
+select derived::base from derived;
+drop table derived;
+drop table base;

+create table p1(ff1 int);
+create table p2(f1 text);
+create function p2text(p2) returns text as 'select $1.f1' language sql;
+create table c1(f3 int) inherits(p1,p2);
+insert into c1 values(123456789, 'hi', 42);
+select p2text(c1.*) from c1;
+drop function p2text(p2);
+drop table c1;
+drop table p2;
+drop table p1;
-select * from d;

On Sun, 2004-12-12 at 18:11 +1100, Neil Conway wrote:
> Attached is a patch adding regression tests for this change.

Applied.

-Neil