greg@turnstep.com writes:
> I guess I was thinking about this:
> http://groups.google.com/groups?selm=10897.953919427%40sss.pgh.pa.us
> "Average width is pretty bogus because the thing really doesn't have
> any idea of the average length of variable-length columns. I'm thinking
> about improving that in the future, but it may not be worth the trouble,
> because the width isn't used for very much."
That comment predated 7.0, which is a long time ago. We now have
statistics about actual average widths of columns, so the estimates are
not nearly as bogus as they used to be. And with the expanded scope for
hash-based query plans in 7.4, I think the estimated size of hash tables
will become an even more interesting tidbit than it is now.
> I also think that it is used that much: if you look at all the EXPLAINS
> that have come across the various lists over the years, very few (if any)
> utilize the "width" in any important way. It's important for computing
> the cost, but I would like to suggest that the extra "noise" should be
> off by default as most people never make use of it, and an EXPLAIN ANALYZE
> is already quite verbose.
But EXPLAIN has always included a lot of info that the man in the street
wouldn't know how to interpret. I don't think making it less complete
is going to help anyone.
regards, tom lane