fails sanity check on H photo cross-sections between high levels

[will]

Following output occurs with r4154 gcc 4.0.1 mac os x 10.5

SanityCheck finds insane H photo cs, n,l = 71, 65, expected, found = 1.430777e-26, 1.505268e-26, error = 5.074198e-02
SanityCheck finds insane H photo cs, n,l = 71, 66, expected, found = 1.969151e-27, 2.076041e-27, error = 5.284771e-02
...
[many similar lines snipped]
...
SanityCheck finds insane H photo cs, n,l = 100, 98, expected, found = 6.195590e-38, 6.745479e-38, error = 8.498343e-02
SanityCheck finds insane H photo cs, n,l = 100, 99, expected, found = 1.242244e-39, 1.355804e-39, error = 8.741905e-02
SanityCheck finds insanity so exiting



           vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv 
          > PROBLEM DISASTER PROBLEM DISASTER.      <
          > Sorry, something bad has happened.      <
          > Please post this on the Cloudy web site <
          > discussion board at www.nublado.org     <
          > Please send all following information:  <
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 


 Cloudy version number is (trunk, r4154M, experimental)
 Cloudy compiled on Aug 20 2010 in OS Apple MacOS using the g++ 40001 compiler. Mode ILP32, denormalized float: T double: T.

The numbers don't look that insane to me. Is this a precision issue?

I will try the same model on 64-bit linux and report back

[will]

model input

[will]

output

[will]

OK, tried latest trunk (r4173) on 64 bit linux box and got exact same error

 Cloudy version number is (trunk, r4173, experimental)
 Cloudy compiled on Aug 26 2010 in OS Linux (AMD64) using the g++ 40403 compiler. Mode LP6
4, denormalized float: T double: T.

Any ideas?

[rporter]

Fixed in r5253.