On Tue, Jan 02, 2007, Seaclear Theory wrote:
>
> I got error when I install amber9. I compile it using g95 and got following
> error mesages:
>
> ......
> ......
> g95 -c -O3 -fno-second-underscore -march=nocona -ffree-form
> -I../../sander -o ../obj/random.o ../obj/random.f
> In file ../obj/random.f:9
>
> FUNCTION xrandom(idum)
> 1
> Warning (139): Value of function 'xrandom' at (1) is never set
> In file ../obj/random.f:19
>
> FUNCTION irandom(idum,imax)
> 1
> Warning (139): Value of function 'irandom' at (1) is never set
> In file ../obj/random.f:9
>
> FUNCTION xrandom(idum)
> 1
> In file ../obj/random.f:44
>
> sum = sum + xrandom(idum)
> 2
> Error: FUNCTION 'xrandom' is of type REAL(4) at (1) and type REAL(8) at (2)
> make[3]: *** [../obj/random.o] Error 1
> make[3]: Leaving directory `/opt/amber/amber9/src/dcqtp/mod'
> make[2]: *** [libdivcon.a] Error 2
> make[2]: Leaving directory `/opt/amber/amber9/src/dcqtp'
> make[1]: *** [divcon] Error 2
> make[1]: Leaving directory `/opt/amber/amber9/src/sander'
> make: *** [serial] Error 2
>
What does g95 --version report? We compile this code with g95 all the time
(on a variety of platforms) and haven't seen that error....
Having said that, the code in random.F90 does look wrong, but as far as I can
see, never used anywhere anyway. You should be able to go ahead and make
xrandom() a double precision function.
I'm cc-ing this to the people in Florida, who should look into this: why do we
have Monte Carlo and parameter optimization code that is never used? Isn't
it pretty dangerous to have bogus functions in random.F90? This seems like it
just asking for future bugs.
...thanks for the report; we'll try to get a bug fix out soon.
...regards...dac
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Jan 07 2007 - 06:07:03 PST
