Andrew Sullivan ajs at crankycanuck.ca
Mon Oct 29 09:53:00 PDT 2007
On Sun, Oct 28, 2007 at 12:10:46PM +0100, hubert depesz lubaczewski wrote:
> i dont have. the problem i see is that bringing the database down to add
> a simple column is not really acceptable.

You don't take the _database_ down, you take the application down.

> why cant slony simply ignore "v" parts of trigger argument?
> 
> after all - insert/update/delete happens trigger *knows* the structure
> of the table the event took place on.

No, it doesn't.  That's exactly the problem.  It looks like it to a
user, but Slony uses the SPI facility to avoid overhead, which means
that it _can't_ know about these changes.  There is work afoot to
make it more possible (see other messages about this), but those
changes won't be backward compatible.  Now that we have reasonably
stable releases that support older versions, we may be in a position
to add backward incompatible features (we did that to 7.3, for
instance, recently).  But up until 8.3, the features you need to
achieve what you want just weren't there anyway.

A

-- 
Andrew Sullivan  | ajs at crankycanuck.ca
In the future this spectacle of the middle classes shocking the avant-
garde will probably become the textbook definition of Postmodernism. 
                --Brad Holland


More information about the Slony1-general mailing list