Re: Table rewrites vs. pending AFTER triggers
От | Gokulakannan Somasundaram |
---|---|
Тема | Re: Table rewrites vs. pending AFTER triggers |
Дата | |
Msg-id | 9362e74e0801030232r1a5781a1l1d20c83a0da02a1@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Table rewrites vs. pending AFTER triggers (Martijn van Oosterhout <kleptog@svana.org>) |
Список | pgsql-hackers |
On Jan 3, 2008 3:53 PM, Martijn van Oosterhout <kleptog@svana.org> wrote:
On Thu, Jan 03, 2008 at 01:08:47PM +0530, Gokulakannan Somasundaram wrote:> Can you please explain, any specific use-case where DDLs are necessaryI don't think they are ever necessary, they're just very very nice. For
> within a transaction?
example:
- You want a new column to appear populated on a table atomically. You
do a BEGIN; add column; update set column=foo; add foreign key; COMMIT
- Installation of external modules can be done atomically, so you don't
end up with half installed contrib modules.
- Principle of Least Surprise. Automatic commit for any reason seems
wrong.
- Temporarily disabling triggers/indexes/constraints, if the system
aborts/crashes, the triggers are reinstated automatically.
- Just general niceity of being able to test schema changes without
immediatly changing the system.
There are many more...
Thanks.. it looks like a good feature...
Gokul.
Gokul.
В списке pgsql-hackers по дате отправления: