Re: Incorrect processing of CREATE TRANSFORM with DDL deparding

On Sat, Jun 27, 2015 at 6:31 AM, Alvaro Herrera wrote:
> Michael Paquier wrote:
>> On Mon, Jun 22, 2015 at 4:25 AM, Alvaro Herrera wrote:
>> > Thanks, pushed that part of it. Looking at the event trigger bits next.
>>
>> Thanks for looking at that. Not sure if that's useful, but attached is
>> a rebased version of the rest.
>
> Pushed the fixes to transform and the new test file, thanks; didn't push
> the other changes for ObjectAddressEq() and the changes in
> ProcessUtilitySlow. These are attached. I don't want to lose that
> stuff: I would like to have an assertion that verifies that the branch
> in the large switch has at least given some thought to command
> collection (precisely to prevent bugs of the type being fixed here.)
> One problem is cases that return InvalidObjectAddress because of IF NOT
> EXISTS, for instance, so it's not just a one-line assert addition.

OK, but actually what does it mean to register an event on an invalid
object. You cannot get a definition from it, right? Hence you cannot
register an event assigned to it.

> I just noticed I neglected to credit for authorship in the commit
> message, sorry about that :-(

Don't worry. That's fine as-is.
--
Michael