Mathematica: TIP: Protecting against computations breaking verbosely because of Null, None, Undefined. And about using Undefined as a return policy.

In a utility package define the following:

isDefNotNull[var_] := !(var===Null) && !(var===None) && !(var===Undefined);

isAllDefNotNull[vars_] := AllTrue[vars,isDefNotNull[#]&];

Then throughout your entire system cleanly test before computing; if the involved values are not defined (in the sense of the above test) return Undefined. For example:

mySafeFunction[x_,y_] := If[isAllDefNotNull[{x,y}],
    x+y,
    Undefined
];

For that simple example, if you did not test first, for x=1. and y=Null you'd just get the relatively harmless partially symbolic:

1.+Null

But for more complex calculations you'd end up with a complete mess. According the the returned strategy advocated here, the returned Undefined is used throughout an application to communicate that not all required information is available. This can be used to communicate progressive value via the user interface. You can even use special styling such as decorating returned Undefined in Orange.

In this example safe unit conversion is performed:

convertSafe[v_,unit_] := If[isDefNotNull[v], UnitConvert[v,unit], Undefined];

You can use the strategy to collect all arguments as a List and test them before passing them to a lower level function using Apply shorthand @@ like this:

service[x_,y_,z_] := ....

safeService[x_,y_,z_] := With[
   {args={x,y,z}},
   If[isAllDefNotNull[args],
     service@@args,
     Undefined
   ] 
]

Of course, if you are using an on-the-fly cascading derived value accessors you can do this:

derived[] := With[
   {args={x[],y[],z[]}},
   If[isAllDefNotNull[args],
     service@@args,
     Undefined
   ] 
]

Diagnostics about why a computation is Undefined can also be included, but that's beyond the scope of this tip.