| On alpha OSF machines
only. The copy of one column of a
bidimensional array into a mono-dimensional array is not working.
The complete bidimensional
array is copied. This is due to an optimisation problem in the
alpha OSF fortran compiler. Reducing the optimisation level from
O4 to
O2 for the routine
kxvcop fixes the problem.
This macro shows the problem:
Vector/create v(2,10) rAppl Data v.dat
1 5
2 7
3 3
4 9
v.dat
Vector/read v v.dat
Vector/print v
Vector/create vbid(10)
Vector/copy v(1) vbid
Vector/print vbid
It gives the output:
PAW > exe a
V(1, 1) = 1
V(2, 1) = 5
V(1, 2) = 2
V(2, 2) = 7
V(1, 3) = 3
V(2, 3) = 3
V(1, 4) = 4
V(2, 4) = 9
V(1, 5) ...
V(2,10) = 0
VBID( 1) = 1
VBID( 2) = 5
VBID( 3) = 2
VBID( 4) = 7
VBID( 5) = 3
VBID( 6) = 3
VBID( 7) = 4
VBID( 8) = 9
VBID( 9) = 0
VBID(10) = 0
PAW >
instead of:
PAW > exe a
V(1, 1) = 1
V(2, 1) = 5
V(1, 2) = 2
V(2, 2) = 7
V(1, 3) = 3
V(2, 3) = 3
V(1, 4) = 4
V(2, 4) = 9
V(1, 5) ...
V(2,10) = 0
VBID( 1) = 1
VBID( 2) = 2
VBID( 3) = 3
VBID( 4) = 4
VBID( 5) ...
VBID(10) = 0
PAW >
|
