From: Tom Lane <tgl@sss.pgh.pa.us>
To: Pavel Stehule <pavel.stehule@gmail.com>
cc: Heikki Linnakangas <hlinnaka@iki.fi>, pgsql-hackers@lists.postgresql.org
Subject: Re: Improving PL/Tcl's error context reports
In-reply-to: 
 <CAFj8pRBTgCHRacAeygUaLCwggARX5Pxqvs+puduiUuRdx9sYHw@mail.gmail.com>
References: <890581.1717609350@sss.pgh.pa.us>
 <974c2531-536a-4223-a264-166872386459@iki.fi>
 <CAFj8pRBTgCHRacAeygUaLCwggARX5Pxqvs+puduiUuRdx9sYHw@mail.gmail.com>
Comments: In-reply-to Pavel Stehule <pavel.stehule@gmail.com>
	message dated "Thu, 04 Jul 2024 19:15:55 +0200"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <26371.1720114586.1@sss.pgh.pa.us>
Content-Transfer-Encoding: 8bit
Date: Thu, 04 Jul 2024 13:36:26 -0400
Message-ID: <26372.1720114586@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/26372.1720114586%40sss.pgh.pa.us>
Precedence: bulk

Pavel Stehule <pavel.stehule@gmail.com> writes:
> čt 4. 7. 2024 v 17:27 odesílatel Heikki Linnakangas <hlinnaka@iki.fi>
> napsal:
>> What happens if you rename a function? I guess the error context will
>> still print the old name, but that's pretty harmless.

> The rename should to generate different tid, so the function will be
> recompiled

Right.  With patch:

regression=# create function bogus() returns int language pltcl as
'return [expr 1 / 0]';
CREATE FUNCTION
regression=# select bogus();
ERROR:  divide by zero
CONTEXT:  while executing
"expr 1 / 0"
    (procedure "__PLTcl_proc_bogus" line 2)
    invoked from within
"__PLTcl_proc_bogus"
in PL/Tcl function "bogus"
regression=# alter function bogus() rename to stillbogus;
ALTER FUNCTION
regression=# select stillbogus();
ERROR:  divide by zero
CONTEXT:  while executing
"expr 1 / 0"
    (procedure "__PLTcl_proc_stillbogus" line 2)
    invoked from within
"__PLTcl_proc_stillbogus"
in PL/Tcl function "stillbogus"

>> Hmm, could we do something with tcl namespaces to allow having two
>> procedures with the same name? E.g. create a separate namespace, based
>> on the OID, for each procedure. I wonder how the stack trace would look
>> like then.

If the namespace depends on the OID then we still have nonreproducible
stack traces, no?  We could maybe make Tcl namespaces that match the
SQL schema names, but that doesn't get us out of the duplication
problem when there are similarly-named functions with different
argument lists.

Another idea is to make the Tcl names include the SQL schema name,
that is __PLTcl_proc_myschema_myfunction.  That avoids needing to
append OIDs when the problem is functions in different schemas,
but it doesn't move the needle for overloaded functions.  On the
whole I feel like that'd add verbosity without buying much.

			regards, tom lane