Re: tablecmds.c/MergeAttributes() cleanup

On 12.07.23 16:29, Peter Eisentraut wrote:
> On 11.07.23 20:17, Alvaro Herrera wrote:
>> I spent a few minutes doing a test merge of this to my branch with NOT
>> NULL changes.  Here's a quick review.
>>
>>> Subject: [PATCH 01/17] Remove obsolete comment about OID support
>>
>> Obvious, trivial.  +1
>>
>>> Subject: [PATCH 02/17] Remove ancient special case code for adding
>>> oid columns
>>
>> LGTM; deletes dead code.
>>
>>> Subject: [PATCH 03/17] Remove ancient special case code for dropping oid
>>>   columns
>>
>> Hmm, interesting.  Yay for more dead code removal.  Didn't verify it.
>
> I have committed these first three.  I'll leave it at that for now.

I have committed a few more patches from this series that were already
agreed upon. The remaining ones are rebased and reordered a bit, attached.

There was some doubt about the patch "Refactor ATExecAddColumn() to use
BuildDescForRelation()" (now 0009), whether it's too clever to build a
fake one-item tuple descriptor. I am working on another patch, which I
hope to post this week, that proposes to replace the use of tuple
descriptors there with a List of something. That would then allow maybe
rewriting this in a less-clever way. That patch incidentally also wants
to move BuildDescForRelation from tupdesc.c to tablecmds.c (patch 0007
here).