Re: DO ... RETURNING

2013/6/10 Stephen Frost <sfrost(at)snowman(dot)net>:
> * Pavel Stehule (pavel(dot)stehule(at)gmail(dot)com) wrote:
>> not too much. Two different concepts in one statement is not good
>> idea.
>
> What are the different concepts..? We already have set returning
> functions, why would set returning anonymous functions be any different?

1. DO as function
2. DO as batch

>
>> What using a cursors as temporary solution?
>
> That only works when you want to just return the results of a table.
> What if you want to construct the data set in the DO block? Okay, fine,
> you could use a temp table, but what if you don't have rights to create
> temporary tables?
>
>> Still I don't like this idea, because you should to support DO
>> RETURNING in other statements - like INSERT INTO DO RETURNING ???
>
> That would certainly be neat, but it doesn't have to be there in the
> first incarnation, or really, ever, if it turns out to be painful to do.
>

this is reason, why I dislike it - It is introduce significant strange
SQL extension

>> What about local temporary functions ??
>
> You can already create temporary functions by simply creating them in
> pg_temp. I'd like to see us add explicit support for them though, but I
> don't see this as related to the DO-RETURNING question.

I don't think we have to introduce a new NON ANSI concept, when is
possible using current feature.

so for me -1

Pavel

>
> Thanks,
>
> Stephen