Peter Geoghegan <pg@bowt.ie> writes:
> I would like to give another 24 hours for anybody to lodge final
> objections to what I've done in this patch. It seems possible that
> there will be concerns about how this might affect backpatching, or
> something like that. This patch goes relatively far in the direction
> of refactoring to make things consistent at the module level -- unlike
> most of the patches, which largely consisted of mechanical adjustments
> that were obviously correct, both locally and at the whole-module level.
Yeah. I'm not much on board with the AHX->A and AH->A changes you made;
those seem extremely invasive and it's not real clear that they add a
lot of value.
I've never thought that the Archive vs. ArchiveHandle separation in
pg_dump was very well thought out. I could perhaps get behind a patch
to eliminate that bit of "abstraction"; but I'd still try to avoid
wholesale changes in local-variable names from it. I don't think that
would buy anything that's worth the back-patching pain. Just accepting
that Archive[Handle] variables might be named either AH or AHX depending
on historical accident does not seem that bad to me. We have lots more
and worse naming inconsistencies in our tree.
regards, tom lane