Hi,
thanks to Antonio Galea, we now have SQL transactions
in the SQLite driver which speeds up "execute" for
large attribute tables:
### EXECUTE, SQLite driver
# TEST 1: geodetic_pts, 13 columns, points=29939
# without transactions, points map
time g.copy vect=geodetic_pts,ggg
real 0m8.076s
user 0m4.135s
sys 0m5.182s
# with transactions, points map
time g.copy vect=geodetic_pts,ggg2
real 0m7.789s
user 0m4.192s
sys 0m5.424s
#####
# TEST 2: streets_wake, 35 columns, lines=49746
# without transactions, lines map
time g.copy vect=streets_wake,ggg2
real 0m22.692s
user 0m11.685s
sys 0m15.315s
# with transactions, lines map
time g.copy vect=geodetic_pts,ggg2
real 0m7.663s
user 0m4.101s
sys 0m4.327s
#####
# TEST 3, soils_wake, 7 columns, areas=47156
# without transactions, area map
time g.copy vect=soils_wake,ggg2
real 0m9.856s
user 0m4.940s
sys 0m5.837s
# with transactions, area map
time g.copy vect=soils_wake,ggg2
real 0m10.208s
user 0m4.986s
sys 0m6.319s
The improvement is (naturally) significant when you have
(a) large attribute table(s) attached to the vector map.
Markus