From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
cc: Pg Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: useless RangeIOData->typiofunc
In-reply-to: <20200304215711.GA8732@alvherre.pgsql>
References: <20200304215711.GA8732@alvherre.pgsql>
Comments: In-reply-to Alvaro Herrera <alvherre@2ndquadrant.com>
	message dated "Wed, 04 Mar 2020 18:57:11 -0300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <8260.1583361293.1@sss.pgh.pa.us>
Date: Wed, 04 Mar 2020 17:34:53 -0500
Message-ID: <8261.1583361293@sss.pgh.pa.us>
Precedence: bulk

Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> I noticed while going over the multirange types patch that it adds a
> pointless typiofunc cached OID to a struct used for I/O functions'
> fn_extra.  It seems to go completely unused, so I checked range types
> (which this was cribbed from) and indeed, it is completely unused there
> either.  My guess is that it was in turn cribbed from array's
> ArrayMetaState, which is considerably more sophisticated; I suspect
> nobody noticed that caching it was pointless.

> Here's a patch to remove it from rangetypes.c.  It doesn't really waste
> much memory anyway, but removing it lessens the cognitive load by one or
> two bits.

Hm, I'm not sure that really lessens the cognitive load any, but
if you do commit this please fix the dangling reference you left
in the nearby comment:

 {
    TypeCacheEntry *typcache;   /* range type's typcache entry */
-   Oid         typiofunc;      /* element type's I/O function */
    Oid         typioparam;     /* element type's I/O parameter */
    FmgrInfo    proc;           /* lookup result for typiofunc */
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^
 } RangeIOData;


			regards, tom lane