op 13-09-2015 01:45 schreef E.Pasma op pasma10 at concepts.nl:
>>
>> On 2015-09-13 12:40 AM, Aurel Wisse wrote:
>>> @R.Smith
>>> I wasn't aware that recursive aggregate queries give wrong answers
>>> in 99%
>>> of all general use cases//...
>>
>> I apologize - I thought Richard had made that much clear early on in
>> the exchange and you were simply not recognizing or accepting the
>> fact, which led to my unnecessarily verbose reply. (Also, 99% might
>> be an exaggerated figure, but the principle remains).
>>
>> Either way, I am happy you have now a query that is even faster to
>> completion than the original one in dispute. Be sure to post any
>> query that takes too long, someone here might have a way of speeding
>> it up.
>>
>> Cheers,
>> Ryan
>>
> Apart from speed also the results are changed. Using drh's securities
> table with 1.000 rows, I get the following line counts:
>
> 999 - aurel's original version is 3.8.9
> 1000 - Luuks non-recursive version, same for my own trial
> 827 -the version with temp table
> 838 - DRH;s version
>
> And the winner is..
>
>
>
> _______________________________________________
> sqlite-users mailing list
> sqlite-users at mailinglists.sqlite.org
> http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Excuse, it is getting late here:
the temp table version returns just one row less then DRH's version.
