Sergey,

Thank for your help.

 SET COLLATE TO "MACHINE"

 a = SECONDS()
  *-- On veut la somme des catégories ( OABCD ) et hospitalisation par type institution ,
  *-- sur une période donnée ( D_debut , d_fin ) avec une extrapolation possible de 8 jours
  *-- sur les HospItalisations.
  SET SAFETY OFF

  SELECT Etats
  INDEX ON ddebut TAG ddebut ADDITIVE
  INDEX ON dfin 	TAG dfin ADDITIVE
  INDEX ON ntype 	TAG ntype ADDITIVE
  INDEX ON nPoscat TAG nPoscat ADDITIVE


  SELECT DatesFiltred
  INDEX ON ntype  TAG ntypeIndex ADDITIVE
  INDEX ON ddate TAG ddate ADDITIVE
  INDEX ON nPoscat TAG oscatIndex ADDITIVE


  SYS(3054,11,"coptimation")

  SELECT DatesFiltred.nPoscat, ;
    Etats.ccat_physiq  ,;
    Etats.ntype , ;
    COUNT( DatesFiltred.ddate ) AS "Ncat" , ;
    Etats.ltransfert  ;
    FROM  DatesFiltred ;
    INNER JOIN amline!Etats ON DatesFiltred.nPoscat = Etats.nPoscat AND DatesFiltred.ntype = Etats.ntype AND ;
    DatesFiltred.ddate BETWEEN Etats.ddebut AND Etats.dfin  ;
    WHERE DatesFiltred.ddate BETW D_debut AND d_fin  ;
    ORDER BY  ltransfert , DatesFiltred.ntype , DatesFiltred.nPoscat ;
    GROUP BY  ltransfert , DatesFiltred.ntype , DatesFiltred.nPoscat ;
    INTO CURSOR Sum_Categ_per_periode && CURSOR tempstat2

  SYS(3054,0)


  CREATE CURSOR TT ( gmemo M )
  INSERT INTO TT ( gmemo ) VALUE( coptimation )
  CLEAR
  ? coptimation


  * WAIT
  B = SECONDS()
  WAIT WINDOW NOWAIT TRANSFORM( B - a , "9999.99999")

Using index tag Ddate to rushmore optimize table datesfiltred
Rushmore optimization level for table datesfiltred: partial
Rushmore optimization level for table etats: none
Joining table etats and table datesfiltred using temp index

No improvement





>Try to add indexes on the JOIN fields 'nposcat' and 'nType' in both tables. You can aslo play with JOIN itself

    FROM  DatesFiltred ;
>    JOIN Etats ON DatesFiltred.nposcat = Etats.nposcat AND DatesFiltred.nType = Etats.nType AND ;
>     DatesFiltred.ddate BETWEEN Etats.ddebut AND Etats.dfin  ;
>* or
>    JOIN Etats ON DatesFiltred.nposcat = Etats.nposcat AND DatesFiltred.nType = Etats.nType ;
>    WHERE DatesFiltred.ddate BETWEEN Etats.ddebut AND Etats.dfin  ;
>

>>Sergey,
>>
>>I make 3 index but Vfp don't want optimze my requery.
>>Why ?
>>
>>
>> SYS(3054,11,"coptimation")
>>
>>

>>  SELECT DatesFiltred.nposcat, ;
>>    Etats.ccat_physiq  ,;
>>    Etats.nType , ;
>>    COUNT( DatesFiltred.ddate ) AS "Ncat" , ;
>>    Etats.ltransfert  ;
>>    FROM  DatesFiltred ,Etats;
>>    WHER DatesFiltred.nposcat = Etats.nposcat AND DatesFiltred.nType = Etats.nType AND ;
>>     DatesFiltred.ddate BETWEEN Etats.ddebut AND Etats.dfin  ;
>>    ORDER BY  ltransfert , DatesFiltred.nType , DatesFiltred.nposcat ;
>>    GROUP BY  ltransfert , DatesFiltred.nType , DatesFiltred.nposcat ;
>>    INTO CURSOR Sum_Categ_per_periode && CURSOR tempstat2
>>
>>  SYS(3054,0)
>>
>>Rushmore optimization level for intermediate result: none
>>Rushmore optimization level for table etats: none
>>Joining table etats and intermediate result using temp index
>>
>>

>
>>>
>>>I don't see anything interesting in that mesage besides incorrect statement that BETWEEN isn't Rushmore optimizable. In fact it is.