In using v2.00/13 I just came across this oddity: the following macro (yes, the
first line is missing a semicolon)
{
Float_t x = 1.0
Float_t y = 2.0;
Float_t z = 3.0;
printf("x = %f\n", x);
printf("z = %f\n", z);
}
produces this output
root [0] .x test.C
x = 0.000000
z = 3.000000
I guess I'd have expected CINT to mark the first line as a syntax error instead
of silently setting x = 0. I tried upping the security level, but there was no
change in the result
root [1] #pragma security level3
root [2] .x test.C
x = 0.000000
z = 3.000000
I poked through the roottalk archives, but I didn't see this discussed (though I
certainly may have missed it). Is this a known behavior of CINT? If so,
doesn't it seem like a somewhat dangerous way to handle a very common error?
Does this come about because CINT doesn't require a semicolon at the command
line?
Actually, if I type the lines of my macro one-by-one at the command line
everything works fine:
root [0] Float_t x = 1.0
root [1] Float_t y = 2.0;
root [2] Float_t z = 3.0;
root [3] printf("x = %f\n",x);
x = 1.000000
root [4] printf("z = %f\n",z);
z = 3.000000
Any guidance on how to convince CINT to complain about my syntax error would be
greatly appreciated.
Thanks,
Dave
--
David Morrison Brookhaven National Laboratory phone: 516-344-5840
Physics Department, Bldg 510 C fax: 516-344-3253
Upton, NY 11973-5000 email: dave@bnl.gov