Hi,
This is product search... the data comes from diffrent dealers and joins to one big database.
In current database request of "%HP%PRINTER" return me 561 record... current table reccount nearly 500 000 records. request duration nearly 10-15 seconds on P-4/2.6Ghz/512 Ram... (Depends from cached previuos query) over network - speed worse..
The task like fulltextsearch.. locate entry of words in string. (not memo.. only character)

The request (UPPER(Description) like "%HP%PRINTER") must returns strings like
"SP/HP Cable USB mini A-B Camera/Drucker"
"HP Papier A4 matt Tintenstrahldrucker"
"DRU Tinte HP Business InkJet 1000 Farbdrucker"

I thinking to use PHDBase.. but not sure about performance when data add into table.. also.. as I understand.. any INSERT/UPDATE/DELETE PHDBase operations need to add code to restruct index..
Thats increase development time... because not there single business object which provide table modification..
At this reason.. I try to find common VFP way to solve this task...

Denis




>Hi Denis
>
>What are you trying to do? How many printer records can you have in this table?
>
>>Hi Again...
>>I try to do next way
>>
>>INDEX ON "1"+UPPER(Description) TAG Test1
>>
>>SELECT * FROM table WHERE "1"+UPPER(Description) like "1%HP%DRUCKER"
>>
>>SYS(3054) show result about Test1 and Deleted Tags using in the Rushmore optimization, but optimization level still is "partial"
>>
>>And... time of exec query realy looks not good....
>>
>>
>>>Hello,
>>>
>>>a LIKE condition which starts with % cannot be optimized.
>>>
>>>Regards
>>>Christian