Re: Rework manipulation and structure of attribute mappings

Hi Michael,

On Mon, Dec 9, 2019 at 11:57 AM Michael Paquier <michael@paquier.xyz> wrote:
> On Fri, Dec 06, 2019 at 06:03:12PM +0900, Amit Langote wrote:
> > Sorry I don't understand this.  Do you mean we should rename the
> > routines left behind in tupconvert.c?  For example,
> > convert_tuples_by_name() doesn't really "convert" tuples, only builds
> > a map needed to do so.  Maybe build_tuple_conversion_map_by_name()
> > would be a more suitable name.
>
> I had no plans to touch this area nor to rename this layer because
> that was a bit out of the original scope of this patch which is to
> remove the confusion and random bets with map lengths.  I see your
> point though and actually a name like what you are suggesting reflects
> better what the routine does in reality. :p

Maybe another day. :)

> So, a couple of hours after looking at the code I am finishing with
> the updated and indented version attached.  What do you think?

Thanks for the updated patch.  I don't have any comments, except that
the text I suggested couple of weeks ago no longer reads clear:

+ * by DDL operating on inheritance and partition trees to convert fully
+ * transformed expression trees from parent rowtype to child rowtype or
+ * vice-versa.

Maybe:

...to adjust the Vars in fully transformed expression trees to bear
output relation's attribute numbers.

Thanks,
Amit