Re: Backend-internal SPI operations - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Backend-internal SPI operations
Date
Msg-id 22049.967667292@sss.pgh.pa.us
Whole thread Raw
In response to Re: Backend-internal SPI operations  ("Mark Hollomon" <mhh@nortelnetworks.com>)
List pgsql-hackers
"Mark Hollomon" <mhh@nortelnetworks.com> writes:
> 2. CREATE RULE ... AS ON SELECT DO INSTEAD ... sets relkind to RELKIND_VIEW
>     and deletes any relation files.

>   q: If we find an index, should we drop it, or complain, or ignore it?
>   q: Should the code check to see if the relation is empty (no valid tuples)?

I think we can ignore indexes.  However, it seems like a wise move to
refuse to convert a nonempty table to view status, *especially* if we
are going to blow away the physical file.  Otherwise mistyping the
relation name in a CREATE RULE could be disastrous (what? you wanted
that data?)
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [BUG] calling lo_creat()
Next
From: Jan Wieck
Date:
Subject: Re: Patch for TNS services