Обсуждение: Documentation on information_ schema columns that does not exist

=?UTF-8?B?Q2zDqW1lbnQgUHLDqXZvc3Q=?= <prevostclement@gmail.com> writes:
> I found out that the following columns are documented (9.1->9.4) but not
> present in the information_schema table of my 9.4 instance (ubuntu):
> * information_schema.foreign_table_options.foreign_server_catalog
> * information_schema.foreign_table_options.foreign_server_name

Yeah, that does seem like a copy-and-pasteo; there should only be 5
columns in the view according to the SQL standard and our code.

Another problem in the same area is that the column types of
foreign_table_schema and foreign_table_name seem to be "name":

# \d *.foreign_table_options
            View "information_schema.foreign_table_options"
        Column         |               Type                | Modifiers
-----------------------+-----------------------------------+-----------
 foreign_table_catalog | information_schema.sql_identifier |
 foreign_table_schema  | name                              |
 foreign_table_name    | name                              |
 option_name           | information_schema.sql_identifier |
 option_value          | information_schema.character_data |

The documentation claims these should be sql_identifier, and that's
what I'd expect in a SQL-standard view ...

            regards, tom lane

I wrote:
> Another problem in the same area is that the column types of
> foreign_table_schema and foreign_table_name seem to be "name":

Further pursuant to that, these are the information_schema columns
that don't seem to have been cast to the spec-required type:

 column_options            | column_name          | name
 column_options            | table_name           | name
 column_options            | table_schema         | name
 foreign_table_options     | foreign_table_name   | name
 foreign_table_options     | foreign_table_schema | name
 foreign_tables            | foreign_table_name   | name
 foreign_tables            | foreign_table_schema | name

            regards, tom lane

On 5/31/15 7:25 PM, Clément Prévost wrote:
> I attached 2 patches: 1 to fix erroneous documentation and the other to
> fix information schema types.

I have fixed these two.

--
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On 5/31/15 8:27 PM, Clément Prévost wrote:
> Here is another patch to fix the name of the column
> information_schema.routines.result_cast_char_set_name that was
> mistakenly named result_cast_character_set_name.

This one is actually named result_cast_character_set_name in the
standard, although that might be a mistake.  Anyway, our documentation
doesn't match our implementation, so one of them has to change.

--
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> On 5/31/15 8:27 PM, Clément Prévost wrote:
>> Here is another patch to fix the name of the column
>> information_schema.routines.result_cast_char_set_name that was
>> mistakenly named result_cast_character_set_name.

> This one is actually named result_cast_character_set_name in the
> standard, although that might be a mistake.  Anyway, our documentation
> doesn't match our implementation, so one of them has to change.

Hm, yeah, I think I vote with the "it's a mistake" camp.  The adjacent
columns are named "result_cast_char_set_whatever", and there are
occurrences of result_cast_char_set_name elsewhere, for example
in the method_specifications view.

            regards, tom lane

On 8/3/16 3:27 PM, Tom Lane wrote:
> Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
>> On 5/31/15 8:27 PM, Clément Prévost wrote:
>>> Here is another patch to fix the name of the column
>>> information_schema.routines.result_cast_char_set_name that was
>>> mistakenly named result_cast_character_set_name.
>
>> This one is actually named result_cast_character_set_name in the
>> standard, although that might be a mistake.  Anyway, our documentation
>> doesn't match our implementation, so one of them has to change.
>
> Hm, yeah, I think I vote with the "it's a mistake" camp.  The adjacent
> columns are named "result_cast_char_set_whatever", and there are
> occurrences of result_cast_char_set_name elsewhere, for example
> in the method_specifications view.

Fixed by changing the column name.

--
Peter Eisentraut              http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services