Subject: Re: Documentation about PL transforms
To: Peter Eisentraut <peter.eisentraut@enterprisedb.com>,
 PostgreSQL Hackers <pgsql-hackers@postgresql.org>
References: <61FDBCFE.3090800@anastigmatix.net>
 <efbdcf7e-9b39-631b-011a-b75cb58e1dd3@enterprisedb.com>
From: Chapman Flack <chap@anastigmatix.net>
Message-ID: <62017DB3.5020900@anastigmatix.net>
Date: Mon, 7 Feb 2022 15:14:43 -0500
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
 Thunderbird/38.7.0
MIME-Version: 1.0
In-Reply-To: <efbdcf7e-9b39-631b-011a-b75cb58e1dd3@enterprisedb.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/62017DB3.5020900%40anastigmatix.net>
Precedence: bulk

On 02/07/22 10:59, Peter Eisentraut wrote:
> On 05.02.22 00:55, Chapman Flack wrote:
>> I'm thinking plhandler.sgml is the only place that really needs to be
>> ...
>> worth mentioning there, though, that it isn't possible to develop
>> transforms for an arbitrary PL unless that PL applies transforms.)
> 
> makes sense
> 
>> (argument_type [, ...]) even though check_transform_function will reject
>> any argument list of length other than 1 or with type other than internal.
> 
> That could be corrected.

I'll work on some doc patches.

>> As long as a PL handler has the sole responsibility for invoking
>> its transforms, I wonder if it would make sense to allow a PL to
>> define what its transforms should look like, maybe replacing
>> check_transform_function with a transform_validator for the PL.
> 
> Maybe.  This kind of thing would mostly be driven what a PL wants in actual
> practice, and then how that could be generalized to many/all PLs.

It has since occurred to me that another benefit of having a
transform_validator per PL would be immediate error reporting
if someone, for whatever reason, tries out CREATE TRANSFORM
for a PL that doesn't grok transforms.

Regards,
-Chap