DevOps, CI/CD and Automation

I am trying to port an extremely large C program from a Sun OS 5.10 system to Linux. The program on the Sun machine was compiled using the Sun C 5.8 compiler. While trying to debug why the results are different on the two systems, I discovered that the code on the Sun machine returns zero when a number is being divided by 0, rather than NAN or inf as I would expect.

For example using DBX, I stepped through the following code and temp was computed to be 0.0,

(dbx) next
stopped in initialize_planet_geometry at line 3125 in file "position_att.c"
3125 temp = v->earth_radius_pole / v->earth_radius_equator;
(dbx) list
3125 temp = v->earth_radius_pole / v->earth_radius_equator;
3126 rpole_2_over_requator_2 = temp * temp;
3127 v->earth_ecc_2 = 1.0 - rpole_2_over_requator_2;
3128
3129 v->requator_2_over_rpole_2 = 1.0 / rpole_2_over_requator_2;
3130
3131 v->earth_flattening = 1.0 - v->earth_radius_pole /
3132 v->earth_radius_equator;
3133
3134 SET_VECTOR( v->earth_rot_vec, 0.0, 0.0, v->earth_rot_rate );
(dbx) print v->earth_radius_equator
v->earth_radius_equator = 0.0
(dbx) print v->earth_radius_pole
v->earth_radius_pole = 0.0
(dbx) next
stopped in initialize_planet_geometry at line 3126 in file "position_att.c"
3126 rpole_2_over_requator_2 = temp * temp;
(dbx) print temp
temp = 0.0

Can anyone explain how this can happen?