Re: Avoid index rebuilds for no-rewrite ALTER TABLE ALTER TYPE

On Thu, Jul 07, 2011 at 02:44:49PM -0400, Robert Haas wrote:
> On Thu, Jul 7, 2011 at 2:43 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> > On Wed, Jul 6, 2011 at 2:50 PM, Noah Misch <noah(at)2ndquadrant(dot)com> wrote:
> >> Drat; fixed in this version.  My local branches contain a large test battery
> >> that I filter out of the diff before posting.  This time, that filter also
> >> removed an essential part of the patch.
> >
> > OK, I'm pretty happy with this version, with the following minor caveats:
> >
> > 1. I still think the documentation changes could use some
> > word-smithing.  If I end up being the one who commits this, I'll take
> > a look at that as part of the commit.
> >
> > 2. I think it would be helpful to add a comment explaining why
> > CheckIndexCompatible() is calling
>
> Woops, sorry. Hit send too soon.
>
> ...why CheckIndexCompatible() is calling ComputeIndexAttrs().

CheckIndexCompatible() calls ComputeIndexAttrs() to resolve the new operator
classes, collations and exclusion operators for each index column. It then
checks those against the existing values for the same. I figured that was
obvious enough, but do you want a new version noting that?

> Rather than committing this immediately, I'm going to mark this "Ready
> for Committer", just in case Tom or someone else wants to look this
> over and weigh in.

Great. Thanks for reviewing.