Re: Eliminating PD_ALL_VISIBLE, take 2

On Mon, 2013-07-01 at 20:59 -0400, Robert Haas wrote:
> One of several relevant emails is at:
>
> http://www.postgresql.org/message-id/51A7473C.6070208@vmware.com
>
> It is definitely possible that I am mixing up two different things.
> But if I am, I don't know what the other one is.

I believe you are mixing up two different things. The patch in the
commitfest now doesn't cause that problem at all.

The thread above is about one proposal in which Andres "basically
suggested treating all visible as frozen". I threw out the idea that my
proposal was not necessarily in conflict with that one, although others
pointed out some problems with combining them.

However, that only matters if Andres's proposal is going to actually
make it in. Heikki also made a very interesting proposal related to
freezing here:

http://www.postgresql.org/message-id/51A7553E.5070601@vmware.com

and that seems compatible with my proposal (which is one of the
advantages you list).

So, if you object because we're moving toward another incompatible
proposal that is more desirable, then I understand that. It can be a bit
frustrating to me though if my proposal is rejected because one of
several proposals is in conflict. (Not that it's necessarily wrong to do
so, but I'm sure you can see how that is frustrating.)

I'll see if I can help out with Heikki's patch. If it starts to look
like it's going to make it, will you drop this particular objection?

Regards,
Jeff Davis