Message-ID: <53f0b7f1-6290-3b99-5040-ac3541df1475@joeconway.com>
Date: Sat, 12 Aug 2023 09:50:25 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.13.0
Subject: Re: CREATE FUNCTION ... SEARCH { DEFAULT | SYSTEM | SESSION }
Content-Language: en-US
From: Joe Conway <mail@joeconway.com>
To: Jeff Davis <pgsql@j-davis.com>, pgsql-hackers@postgresql.org
References: <2710f56add351a1ed553efb677408e51b060e67c.camel@j-davis.com>
 <0e51f9e2-9272-ccb3-9219-39d7695f7c50@joeconway.com>
In-Reply-To: <0e51f9e2-9272-ccb3-9219-39d7695f7c50@joeconway.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: <https://www.postgresql.org/message-id/53f0b7f1-6290-3b99-5040-ac3541df1475%40joeconway.com>
Precedence: bulk

On 8/12/23 09:15, Joe Conway wrote:
> On 8/11/23 22:35, Jeff Davis wrote:
>> 2. We can more accurately serve the user's intent. For instance, the
>> safe search_path of "pg_catalog, pg_temp" is arcane and seems to be
>> there just because we don't have a way to specify that pg_temp be
>> excluded entirely. But perhaps in the future we *do* want to exclude
>> pg_temp entirely. Knowing that the user just wants "SEARCH SYSTEM"
>> allows us some freedom to do that.
> 
> Personally I think having pg_temp in the SYSTEM search path makes sense
> for temp tables, but I find it easy to forget that functions can be
> created by unprivileged users in pg_temp, and therefore having pg_temp
> in the search path for functions is dangerous.

Hmm, I guess I was too hasty -- seems we have some magic related to this 
already.

-- 
Joe Conway
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com