#2 Potential division-by-zero at spline.c

[Changgong Lee]

Dear authors,
There exists a potential division-by-zero at the function hermite3_interp in libmorph/spline.c

...
281      h   = x[si+1] - x[si];
282      h_2 = h*h;
...
338      A = (y[si+1] - y[si] - h*d[si])/h_2;
339      B = (d[si+1] - d[si] - 2.0*h*A)/h_2;

, when arbitary values are given to x[si+1] and x[si] from
DrawMeshes function in diw_map.c

347      diw_map_t * diw_mapP = diw_map_of_widget(widget)

while
diw_map.c in diw_map_of_widget returns

302      global_diw_map[indx].mesh_dst.y = global_diw_map[0].mesh_dst.y;
...
317      dmP = &global_diw_map[indx];
...
320      retrun dmP

and global_diw_map is read from
initialize_application function in xmorph.c

432      meshRead(&global_diw_map[0].mesh_dst, dstmesh)

A simple possible fix suggestion would be adding additional if statements.
For example,

+        if (!h_2) return 0;
338      A = (y[si+1] - y[si] - h*d[si])/h_2;

[Michael J. Gourlay]

Yeah, your suggestion looks legit. Thanks!

I haven't touched this code in a few decades. I'm glad you're finding some utility from it!