Re: about truncate - Mailing list pgsql-hackers

From Peter Eisentraut
Subject Re: about truncate
Date
Msg-id 496755BF.6000208@gmx.net
Whole thread Raw
In response to Re: about truncate  (David Fetter <david@fetter.org>)
Responses Re: about truncate  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
David Fetter wrote:
> On Thu, Jan 08, 2009 at 02:39:52PM +0200, Peter Eisentraut wrote:
>> David Fetter wrote:
>>> +1 for adding recursion to GRANT/REVOKE :)
>> This area is under SQL standard control, so we can't really invent our  
>> own behavior.
>>
>> Consider the following:
>>
>> CREATE TABLE persons (name, email);
>> CREATE TABLE employees (grade, salary) INHERITS (persons);
>>
>> GRANT SELECT ON persons TO allstaff;  -- ???
>> GRANT SELECT ON employees TO managers;
>>
>> What you want in practice is that allstaff can read only those columns  
>> of employees that come from the persons table.  Both recursive and  
>> nonrecursive GRANT do the wrong thing here.
> 
> What *would* do the right thing here, or would anything?

I think we don't need GRANT to be recursive, but instead the permission 
checks at runtime should allow

SELECT * FROM persons;

to succeed even if there are no permissions on "employees".

But only on the columns of "persons" and only if actually queried 
through "persons".

Needs a more detailed analysis, but that is how I imagine it ought to work.


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Buffer pool statistics in Explain Analyze
Next
From: Tom Lane
Date:
Subject: Re: about truncate