# 2007, M. Spiekermann
#
# A SECONDO-Script which creates synthesized data used for
# experiments with correlated predicates

let ten = [const rel(tuple([A: int, B: int, C: int, D: int])) 
  value ( (0 1 2 3)
          (1 2 3 4)
          (2 3 4 5)
          (3 4 5 6)
          (4 5 6 7)
          (5 6 7 8)
          (6 7 8 9)
          (7 8 9 0)
          (8 9 0 1)
          (9 0 1 2) )];


#create table thousand
#  as select (A.a) * 100 + (B.a) * 10 + C.a as a
#  from ten as A, ten as B, ten as C order by a asc;
#
let thousand = ten feed {A} ten feed {B} ten feed {C} product product extend[No: (.A_A * 100) + (.A_B * 10) + .A_C] project[No] sortby[No asc] consume;


#
#-- Create two objects with one million tuples
#create table m1
#  as select (T.a % 1000) as a, ((T.a + 1) % 1000) as b,
#         ((T.a + 2) % 1000) as c, ((T.a + 3) % 1000) as d
#  from thousand, thousand as T;



let m1 = thousand feed {R} thousand feed {T} product 
 extend[A: .No_T mod 1000] 
  extend[B: (.No_T + 1) mod 1000] 
   extend[C: (.No_T + 2) mod 1000] 
    extend[D: (.No_T + 3) mod 1000] 
project[A, B, C, D] 
consume;

let m2 = m1;

let m1_A = m1 createbtree[A];

# Queries:
# sql select count(*) from [m1 as r, m2 as s] where [r:a = s:a, s:a=0, s:b=1]
#
# Nested Loop (9.43 s)
# query m1_A m1 exactmatch[0] consume 
#    within[fun( x: ANY ) m2 feed filter[.A = 0] {s} loopjoin[x feed] count]
#
# Index Nested Loop (47.66 s)
# m2  feed {s}  filter[(.B_s = 1)]  filter[(.A_s = 0)]  loopjoin[m1_A m1  exactmatch[.A_s] {r} ]  count
#
# Adaptive (hj = 20.09 s) => Selectivity estimation error 100% !!!
#
#
#secondo('query m2  pfeed[100]  puse[. {s}  filter[(.B_s = 1)]  filter[(.A_s = 0)] ] m1 pjoin1[1; symj: . ..  feed  head[1000]  {1} {r}  head[500]  {2} symmjoin[(.A_s = ..A_r)] {4}, hj: . ..  feed {r}  {3} hashjoin[A_s, A_r, 9997] , smj: . ..  feed {r}  {3} sortmergejoin[A_s, A_r] , ilj: .  loopjoin[m1_A ..  exactmatch[.A_s] {r}  {3} ] ]  pdelete  count')