Re: Seems to be impossible to set a NULL search_path
| От | David G. Johnston |
|---|---|
| Тема | Re: Seems to be impossible to set a NULL search_path |
| Дата | |
| Msg-id | CAKFQuwYaYEe0K+Cn1nOZWvfMgAFtE+9iM99h2eiH92YtB8K-jQ@mail.gmail.com обсуждение исходный текст |
| Ответ на | Re: Seems to be impossible to set a NULL search_path (Bryn Llewellyn <bryn@yugabyte.com>) |
| Ответы |
Re: Seems to be impossible to set a NULL search_path
|
| Список | pgsql-general |
On Wed, Jul 6, 2022 at 1:47 PM Bryn Llewellyn <bryn@yugabyte.com> wrote:
It seems that the wording is wrong here:« The value for search_path must be a comma-separated list of schema names. »It's schema identifiers—and not schema names. Yes?
To add further clarity (or confusion) there is also set_config(...)
postgres=# select set_config('search_path','a,"b",testSchema',false);
set_config------------------
a,"b",testSchema
(1 row)
postgres=# select current_schemas(true);
current_schemas
-------------------------
{pg_catalog,testschema}
(1 row)
As for "schema identifiers" vs. "schema names" - they both seem equally wrong. The list can very much contain sequences of characters that when interpreted as an identifier and looked for in the pg_namespace catalog do not find a matching entry and are therefore by definition not the name of any known schema in that database.
Besides, I hazard to guess how many times we write "table name" and "column name" in the documentation when your argument is that "table identifier" and "column identifier" is the correct choice. No, rather "name" and "identifier" in the context of database objects are known to mean the same thing - the alphabetic name of the object.
David J.
В списке pgsql-general по дате отправления: