int integers are allowed but not properly cast

[nanthony21]

If the user passes an array of ints into nlp.create then the c extension will improperly interpret this data as floats without giving any indication of an error. For example numpy.array([1,1,1,1]) become {0.0,0.0,0.0,0.0} and numpy.array([-1,-1,-1,-1]) become {-inf,-inf,-inf,-inf}

[xuy]

That's very interesting. This plus the PR I just merged made me think that a more proper fix for this kind of problem is to wrap a Python class around nlp, which can do all kinds of sanity checks in Python, before going to the C level where you'd need to call various numpy APIs via its C interface.

[nanthony21]

That seems like a good idea. Even now that I know this issue can happen I ran into it again when my eval_h function was occasionally returning an int type array.