--- "Mrs. Brisby" <[EMAIL PROTECTED]> wrote:
> That all said, it seems like this problem is already solved- SQLite does
> the right thing after ANALYZE is called. Perhaps it wouldn't be too
> difficult to update the statistics that ANALYZE collects in after
> COMMIT, or perhaps after a COMMIT that
> People who visit a website that looks like junk _NEVER_ say "oh my web
> browser is being a piece of shit."
...unless they're the same people who designed it ;)
e DBMS's, but at least it would be possible.
Which brings me back to PostgreSQL - which was commented as not having a
special pragma for this but instead a sly little hack identical to the
one SQLite is using. People who target PostgreSQL will already have had
the fix.
> -Original Mes
-Original Message-
From: Mrs. Brisby [mailto:[EMAIL PROTECTED]
Sent: Sunday, 11 September, 2005 10:07 PM
To: sqlite-users@sqlite.org
Subject: RE: [sqlite] CROSS keyword disables certain join optimizations
On Sat, 2005-09-10 at 21:38 -0400, Ned Batchelder wrote:
> Rather than overload an ex
On Sat, 2005-09-10 at 21:38 -0400, Ned Batchelder wrote:
> Rather than overload an existing SQL keyword, would it be possible to
> provide pragmas to control the optimizer? Assigning meanings to particular
> combinations of SQL queries won't scale as the number of optimizer controls
> grows.
I do
On Sat, 2005-09-10 at 19:25 -0400, D. Richard Hipp wrote:
> Beginning with SQLite 3.2.3, the query optimizer has had
> the ability to reorder tables in the FROM clause if it thinks
> that doing so will make the query run faster. This has caused
> a few problems for some folks. To ameliorate those
I vote to not control manual table join order with PRAGMAs, as
PRAGMAs are too course-grained to be of any practical use in
complicated queries. For example, what happens if you want to
force a join on just a couple of tables in the FROM clause,
but not others? It is not possible without an a set o
t; From: Kervin L. Pierre [mailto:[EMAIL PROTECTED]
> Sent: Saturday, 10 September, 2005 9:08 PM
> To: sqlite-users@sqlite.org
> Subject: Re: [sqlite] CROSS keyword disables certain join optimizations
>
> Darren Duncan wrote:
> > At 7:25 PM -0400 9/10/05, D. Richard Hipp wro
On Sat, Sep 10, 2005 at 07:25:48PM -0400, D. Richard Hipp wrote:
> difficulties, I have now modified the query optimizer so
> that it will no longer reorder tables in a join if the join
> is constructed using the CROSS keyword. For additional
This is a one-off to control one particular feature (
To: sqlite-users@sqlite.org
Subject: Re: [sqlite] CROSS keyword disables certain join optimizations
Darren Duncan wrote:
> At 7:25 PM -0400 9/10/05, D. Richard Hipp wrote:
>
> Well, that's fine as long as CROSS still continues to mean and do what
> it has always meant, which is t
Darren Duncan wrote:
At 7:25 PM -0400 9/10/05, D. Richard Hipp wrote:
Well, that's fine as long as CROSS still continues to mean and do what
it has always meant, which is that you explicitly want the result set of
If I understand the issue correctly, it does.
"FROM a, b" is usually equivale
At 7:25 PM -0400 9/10/05, D. Richard Hipp wrote:
The rational behind using the CROSS keyword to disable an
optimization is that the CROSS keyword is perfectly
valid SQL syntax but nobody ever uses it so I figured
we can put it to use to help control the optimizer without
creating any incompatibil
12 matches
Mail list logo
