On 07/11/15 09:59, Nathan Wagner wrote:
[...]
My day to make a fool of myself in public I guess. You're right of
course. I can only plead distraction by having too many projects in
mind at once and not focusing properly. Sorry for taking up your time
on things I should have checked better.
[..
On Fri, Nov 06, 2015 at 02:16:41PM -0500, Tom Lane wrote:
> Uh, what? It's not by any means turned off by default.
>
> postgres=# select name,setting from pg_settings where name like '%geqo%';
> name | setting
> -+-
> geqo| on
[snip]
Nathan Wagner writes:
> On Fri, Nov 06, 2015 at 11:45:38AM -0500, Tom Lane wrote:
>> (There's a fair amount of dead code in /geqo/, which I've never had
>> the energy to clean up, but maybe we should do that sometime. It
>> seems unlikely that anyone will ever be interested in experimenting
>> wi
On Fri, Nov 06, 2015 at 11:45:38AM -0500, Tom Lane wrote:
> However, really the whole argument is moot, because I notice that
> geqo_mutation() is only called in the "#ifdef CX" code path, which
> we don't use.
I suppose someone could turn it on via a compiler define.
> So there's little point i
On Fri, Nov 06, 2015 at 11:19:00AM -0500, Tom Lane wrote:
> Nathan Wagner writes:
> > I see you committed a modified version of my patch in commit
> > 59464bd6f928ad0da30502cbe9b54baec9ca2c69.
>
> > You changed the tour[0] to be hardcoded to 1, but it should be any
> > of the possible gene number
Nathan Wagner writes:
> On Wed, Nov 04, 2015 at 12:51:52PM -0500, Tom Lane wrote:
>> I'm not very impressed with the first patch: it might save a few
>> geqo_randint() calls, but it seems to do so at the price of making the
>> swap choices less random --- for instance it sure looks to me like the
Nathan Wagner writes:
> I see you committed a modified version of my patch in commit
> 59464bd6f928ad0da30502cbe9b54baec9ca2c69.
> You changed the tour[0] to be hardcoded to 1, but it should be any of
> the possible gene numbers from 0 to remainder.
How so? The intent is to replace the first it
On Wed, Nov 04, 2015 at 12:51:52PM -0500, Tom Lane wrote:
> As for the second part, I had to look up Fisher-Yates ;-) but after
> having read Wikipedia's entry about it I think this is a good change.
> The code's shorter and more efficient, and it should mathematically
> provide an equally-unbiase
On Wed, Nov 04, 2015 at 12:51:52PM -0500, Tom Lane wrote:
> Nathan Wagner writes:
> > I have two patches to make the geqo initialization and mutation
> > slightly better.
>
> > The first adjusts the mutation swaps to avoid having to re-pick
> > ties. The second changes the initialization and shu
Alvaro Herrera writes:
> Tom Lane wrote:
>> Having said that, though, I believe that it's also probably a
>> *different* initial shuffle, which may well mean that GEQO gives
>> different plans in some cases. That doesn't bother me as long as
>> we only make the change in HEAD, but does anyone wan
Tom Lane wrote:
> Having said that, though, I believe that it's also probably a
> *different* initial shuffle, which may well mean that GEQO gives
> different plans in some cases. That doesn't bother me as long as
> we only make the change in HEAD, but does anyone want to complain?
Uh, do we pro
Nathan Wagner writes:
> I have two patches to make the geqo initialization and mutation
> slightly better.
> The first adjusts the mutation swaps to avoid having to re-pick
> ties. The second changes the initialization and shuffling algorithm
> for the gene array to use an in-place Fisher-Yates
On 8 September 2015 at 14:44, Nathan Wagner wrote:
> I have two patches to make the geqo initialization and mutation
> slightly better.
>
> The first adjusts the mutation swaps to avoid having to re-pick
> ties. The second changes the initialization and shuffling algorithm
> for the gene array t
I have two patches to make the geqo initialization and mutation
slightly better.
The first adjusts the mutation swaps to avoid having to re-pick
ties. The second changes the initialization and shuffling algorithm
for the gene array to use an in-place Fisher-Yates shuffling
algorithm.
Diffs again
14 matches
Mail list logo
