Re: [PATCH] Unremovable tuple monitoring

On Thu, Nov 17, 2011 at 5:49 PM, Royce Ausburn <royce(dot)ml(at)inomial(dot)com> wrote:
> Thanks for the discussion so far all.  Would it be worthwhile to make another patch that addresses the points from Yeb's reviews?  It's not sounding like this unremovable tuple count is something that postgres wants, but I'm happy to keep the patch up to scratch if we're still not sure.

One question to ask yourself at this point in the process is whether
*you* still think the feature is useful. I'm fairly persuaded by
Tom's point that the value monitored by this patch will change so
quickly that it won't be useful to have VACUUM store it; it'll be
obsolete by the time you look at the numbers. If you are also
persuaded by that argument, then clearly it's time to throw in the
towel!

But let's suppose you're NOT persuaded by that argument and you still
want the feature. In that case, don't just wait for someone else to
stick up for the feature; tell us why you still think it's a good
idea. Make a case for what you want. People here are usually
receptive to good ideas, well-presented.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company