Обсуждение: Bug in CHECK constraints statement reverse engineering.

Поиск
Список
Период
Сортировка

Bug in CHECK constraints statement reverse engineering.

От
Ivan
Дата:
Hello,

PgAdmin 1.3.0 (Apr 24 2005)
Wrong CHECK reverse engineering.

In PostgreSQL documentation i found:

----------------------------------------------------------------------
41.10. pg_constraint
Note
consrc is not updated when referenced objects change; for example,
it won't track renaming of columns. Rather than relying on this field,
it's best to use pg_get_constraintdef() to extract the definition
of a check constraint.
----------------------------------------------------------------------

Here the script to demonstraint this behaviour:

-- First create function and constraints on table field and domain
CREATE OR REPLACE FUNCTION "Check_IntegerGreaterZero"(int4) RETURNS bool AS
$BODY$ select $1 > 0;
$BODY$ LANGUAGE 'sql' VOLATILE SECURITY DEFINER;

CREATE TABLE "TestTable"
( test int4, CONSTRAINT "CHK_TestTable_Test" CHECK ("Check_IntegerGreaterZero"(test))
) 
WITHOUT OIDS;

CREATE DOMAIN "TestDomain" AS int4  CONSTRAINT TestDomain_check CHECK "Check_IntegerGreaterZero"(VALUE);

-- Then just rename function
ALTER FUNCTION "Check_IntegerGreaterZero"(int4) RENAME TO "Check_IGZ";

--Watch pg_constraint and result of pg_get_constraintdef()
select conname, consrc, pg_get_constraintdef(pg_constraint.oid) from pg_constraint


conname                         |consrc                            |pg_get_constraintdef

--------------------------------------------------------------------------------------------------------------------------------
cardinal_number_domain_check    |(VALUE >= 0)                      |CHECK ((VALUE>= 0))
CHK_TestTable_Test              |"Check_IntegerGreaterZero"(test)  |CHECK("Check_IGZ"(test))
TestDomain_check                |"Check_IntegerGreaterZero"(VALUE) |CHECK("Check_IGZ"(VALUE))


I suppose that PgAdmin shows values from consrc.


Another feature request. The function described above ("Check_IntegerGreaterZero") in PgAdmin
right bottom pane shows as
-------------------------------------------------------------------------------------

-- Function: Check_IntegerGreaterZero(in_Value int4)

-- DROP FUNCTION "Check_IntegerGreaterZero"(int4);

CREATE OR REPLACE FUNCTION "Check_IntegerGreaterZero"(int4) RETURNS bool AS
$BODY$ select $1 > 0;
$BODY$ LANGUAGE 'sql' VOLATILE SECURITY DEFINER;
ALTER FUNCTION "Check_IntegerGreaterZero"(int4) OWNER TO postgres;

-------------------------------------------------------------------------------------

It will be convenient for me if the first line will be

-- Function: "Check_IntegerGreaterZero"("in_Value" int4)

- quoted function name and argument names. It is useful for copy / paste purposes :)


Thank you for support.

-- 
Best regards,Ivan                          mailto:Ivan-Sun1@mail.ru



Re: Bug in CHECK constraints statement reverse engineering.

От
"Dave Page"
Дата:

> -----Original Message-----
> From: pgadmin-support-owner@postgresql.org
> [mailto:pgadmin-support-owner@postgresql.org] On Behalf Of Ivan
> Sent: 19 May 2005 16:37
> To: pgadmin-support@postgresql.org
> Subject: [pgadmin-support] Bug in CHECK constraints statement
> reverse engineering.
>
> Hello,
>
> PgAdmin 1.3.0 (Apr 24 2005)
> Wrong CHECK reverse engineering.

Hi,

pgAdmin does do this correctly. In order to run at a reasonable speed,
pgAdmin caches details of objects read from the database, rather than
running queries every time you select one. If you rename an object such
as a function, it doesn't always know that that action may cause a
property of another object to be changed, thus pgAdmin may continue to
show the old definition.

To force a reload, right-click a node in the treeview and select the
'Refresh' option.

>
> It will be convenient for me if the first line will be
>
> -- Function: "Check_IntegerGreaterZero"("in_Value" int4)
>
> - quoted function name and argument names. It is useful for
> copy / paste purposes :)

Thanks, fix commited.

Regards, Dave


Re: Bug in CHECK constraints statement reverse engineering.

От
Ivan
Дата:
Hello Dave,

DP> Hi,

DP> pgAdmin does do this correctly. In order to run at a reasonable speed,
DP> pgAdmin caches details of objects read from the database, rather than
DP> running queries every time you select one. If you rename an object such
DP> as a function, it doesn't always know that that action may cause a
DP> property of another object to be changed, thus pgAdmin may continue to
DP> show the old definition. 

DP> To force a reload, right-click a node in the treeview and select the
DP> 'Refresh' option.

You were right on a half - pgAdmin do this correctly for tables check
constraints, but for domain definition 'Refresh' and even closing
pgAdmin and restarting postmaster doesn't help :) -
I see:

CREATE DOMAIN "TestDomain" AS int4  CONSTRAINT TestDomain_check CHECK "Check_IntegerGreaterZero"(VALUE);

though function were renamed.

By the way it will be great to add quoting of domain's constraint name
in the definition pane (right bottom).

Thank you for support.

-- 
Best regards,Ivan                            mailto:Ivan-Sun1@mail.ru



Re: Bug in CHECK constraints statement reverse engineering.

От
"Dave Page"
Дата:

> -----Original Message-----
> From: Ivan [mailto:Ivan-Sun1@mail.ru]
> Sent: 20 May 2005 10:01
> To: Dave Page
> Cc: pgadmin-support@postgresql.org
> Subject: Re[2]: [pgadmin-support] Bug in CHECK constraints
> statement reverse engineering.
>
> You were right on a half - pgAdmin do this correctly for tables check
> constraints, but for domain definition 'Refresh' and even closing
> pgAdmin and restarting postmaster doesn't help :) -

Ack, sorry - missed that. Fixed in SVN.

Regards, Dave