How to track down doubloons in valentina ?

[Ruslan Zasukhin]

On 9/29/04 9:17 PM, "olivier" <vidal_olivier at yahoo.fr> wrote:

Hi Oliver,

>> It looks to be very hard for execution.
>> It needs to compare each row with each row AFTER it.
>> If you have 100K records, then you get 50K * 100K comparisons...if I
>> count
>> correctly.
> 
> It is a big problem.
> Effectively it asks for many resources.

> It is for it that I hoped that this type of function shall be optimized
> in Valentina.

> I am amazed. It is all the same one of the main purposes of a database
> to verify doubloons and to compare a recording with the other
> recordings of the base ?

No.

There general operation is GROUP BY, DISTINCT,
Which can compare order remove THE SAME values.

Not-exact match is not general operation of dbms


>> I think you need or write some own smart alg
>> May be do some data prepare to optimize this task.
>> 
> 
> I am going to think about it.
> I hope that I am going to find a solution otherwise my program is
> damned!

> I have already tried the basic method: a realbasic loop. it is slow,
> far too much slow.

> I am going to try to find a solution.
> You see a solution?

Look.

By raw force this is incredible by resources task.

I think you need search for special algorithm.
It must do some massage of data. For example sort words in the field.
May be this will help somehow.

May be you need produce table of all and each word and
In the main table produce that field as combination of numbers (recIDs)

Hmm, may be this is an idea.

Let you have 

    1 word1
    2 word2  

Then in your main table you have

    word1 word2     1 + 2       SORT        1 2
    word2 word1     2 + 1       SORT        1 2


I think you need search inet for ideas.



-- 
Best regards,
Ruslan Zasukhin      [ I feel the need...the need for speed ]
-------------------------------------------------------------
e-mail: ruslan at paradigmasoft.com
web: http://www.paradigmasoft.com

To subscribe to the Valentina mail list go to:
http://lists.macserve.net/mailman/listinfo/valentina
-------------------------------------------------------------