From: Tom Lane <tgl@sss.pgh.pa.us>
To: myzhen <zhenmingyang@yeah.net>
cc: pgsql-bugs@postgresql.org
Subject: Re: coredump of language internal function
In-reply-to: <a533370.bb1.1996fc78514.Coremail.zhenmingyang@yeah.net>
References: <a533370.bb1.1996fc78514.Coremail.zhenmingyang@yeah.net>
Comments: In-reply-to myzhen <zhenmingyang@yeah.net>
	message dated "Mon, 22 Sep 2025 12:56:10 +0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2811275.1758517539.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 22 Sep 2025 01:05:39 -0400
Message-ID: <2811276.1758517539@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2811276.1758517539%40sss.pgh.pa.us>
Precedence: bulk

myzhen <zhenmingyang@yeah.net> writes:
> When creating a function using language internal, if the parameters or f=
unction attributes of the alias function are different from those of the i=
nternal function, they are not checked during creation and may cause a cor=
dump when called.

We have always considered that this is programmer error.

> I didn't find this mentioned in the manual, so I'm not sure if this is a=
 problem, but it seems like it might be better to do a check when creating=
 the function.

If you have an idea about how such a check could be done, it would
be worth discussing that.  But it seems pretty unlikely to me that
there is any portable way to identify externally what data types
a compiled function is expecting to receive.

			regards, tom lane