On Tue, Jan 23, 2007 at 10:12:13 -0500,
Brandon Aiken <BAiken@winemantech.com> wrote:
> Out of curiosity, has the COUNT(*) with no WHERE clause slowness been
> fixed in 8.x? Or is it still an issue of "there's no solution that
> won't harm aggregates with WHERE clauses"?
Probably not in the sense that you mean.
The underlying problem is that in MVCC there is no single global answer
to the question and the pain of maintaining the mutliple answers outweighs
the cost of doing so in normal usage.
People that need to run count(*) queries a lot may want to make a different
trade off and some ways of maintaining counts are covered in the archives.