Date: Fri, 12 Jul 2019 17:59:56 +0200 (CEST)
From: Fabien COELHO <coelho@cri.ensmp.fr>
To: Tom Lane <tgl@sss.pgh.pa.us>
cc: Daniel Gustafsson <daniel@yesql.se>, pgsql-docs@lists.postgresql.org,
    Michael Paquier <michael@paquier.xyz>
Subject: Re: SPITupleTable members missing in docs
In-Reply-To: <4584.1562943879@sss.pgh.pa.us>
Message-ID: <alpine.DEB.2.21.1907121748290.8895@lancre>
References: <0D19F836-B743-4340-B6A2-F148CA3DD1F0@yesql.se>
 <alpine.DEB.2.21.1907121649180.8895@lancre> <4584.1562943879@sss.pgh.pa.us>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Precedence: bulk


Hello,

>> To take into account Tom's comment, I'd suggest a middle ground by 
>> commenting a public and private part explicitely in the struct,
>>    typedef struct {
>>      /* PUBLIC members to be used by callers ... */
>>      /* PRIVATE members, not intended for external usage ... */
>>    } ... ;
>
> One problem is that the members we've retroactively decided are "public"
> are in the middle of the struct :-(.

Argh, I did not notice this tiny but relevant detail.

> But it occurs to me that there's no good reason we couldn't re-order the
> members, as long as we only do so on HEAD and not in released versions.
> That would make it a bit less inconsistent and easier to add labels
> such as you suggest.

Indeed. SPI-dependent extensions are likely recompiled between major 
version, so a reordering should not cause significant problems.

This mean that a simple doc patch is turned into a code patch.

-- 
Fabien.