Re: pgsql: Remove IS_AF_UNIX macro - Mailing list pgsql-committers

From Peter Eisentraut
Subject Re: pgsql: Remove IS_AF_UNIX macro
Date
Msg-id 2c53ab2c-e14d-eb00-554e-5520bdbc8d3c@enterprisedb.com
Whole thread Raw
In response to Re: pgsql: Remove IS_AF_UNIX macro  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: pgsql: Remove IS_AF_UNIX macro
List pgsql-committers
On 15.02.22 16:41, Tom Lane wrote:
> Peter Eisentraut <peter@eisentraut.org> writes:
>> Remove IS_AF_UNIX macro
>> The AF_UNIX macro was being used unprotected by HAVE_UNIX_SOCKETS,
>> apparently since 2008.
> 
> I hadn't looked closely at this patch, but are you referring to
> this bit in ip.h?
> 
> #ifdef    HAVE_UNIX_SOCKETS
> #define IS_AF_UNIX(fam) ((fam) == AF_UNIX)
> #else
> #define IS_AF_UNIX(fam) (0)
> #endif
> 
> That's by no means "unprotected": we will not try to reference
> AF_UNIX unless HAVE_UNIX_SOCKETS is set.

In src/backend/utils/adt/pgstatfuncs.c there is a use of AF_UNIX that 
has been there unprotected by any #ifdef since about 2008.

> We might as well just nuke
> all the HAVE_UNIX_SOCKETS conditional compilation if we let
> this stand.
> 
> (Now, maybe we should indeed do that.  I don't have much
> interest in the possibility that we'll worry about such
> platforms in future.)

Maybe/probably.  But there is a difference between platforms having the 
AF_UNIX symbol (which is required by POSIX unconditionally) and 
platforms actually having Unix sockets or not (which might require 
different default configurations or run-time behavior), so the two 
questions are not that closely connected.



pgsql-committers by date:

Previous
From: John Naylor
Date:
Subject: pgsql: Add missing TYPEALIGN macros
Next
From: Tom Lane
Date:
Subject: Re: pgsql: Remove IS_AF_UNIX macro