Misuse of a variables value?

Question

I came across an instance where a solution to a particular problem was to use a variable whose value when zero or above meant the system would use that value in a calculation but when less than zero would indicate that the value should not be used at all.

My initial thought was that I didn't like the multipurpose use of the value of the variable: a.) as a range to be using in a formula; b.) as a form of control logic.

What is this kind of misuse of a variable called? Meta-'something' or is there a classic antipattern that this fits?

Sort of feels like when a database field is set to null to represent not using a value and if it's not null then use the value in that field.

Update: An example would be that if a variable's value is > 0 I would use the value if it's <= 0 then I would not use the value and decided to perform some other logic.

Answer 1

Values such as these are often called "distinguished values". By far the most common distinguished value is null for reference types. A close second is the use of distinguished values to indicate unusual conditions (e.g. error return codes or search failures).

The problem with distinguished values is that all client code must be aware of the existence of such values and their associated semantics. In practical terms, this usually means that some kind of conditional logic must be wrapped around each call site that obtains such a value. It is far too easy to forget to add that logic, obtaining incorrect results. It also promotes copy-and-paste code as the boilerplate code required to deal with the distinguished values is often very similar throughout the application but difficult to encapsulate.

Common alternatives to the use of distinguished values are exceptions, or distinctly typed values that cannot be accidentally confused with one another (e.g. Maybe or Option types).

Having said all that, distinguished values may still play a valuable role in environments with extremely tight memory availability or other stringent performance constraints.

Answer 2

I don't think what your describing is a pure magic number, but it's kind of close. It's similar to the situation in pre-.NET 2.0 where you'd use Int32.MinValue to indicate a null value. .NET 2.0 introduced Nullable and kind of alleviated this issue.

So you're describing the use of a variable who's value really means something other than it's value -- -1 means essentially the same as the use of Int32.MinValue as I described above.

I'd call it a magic number.

Hope this helps.

Answer 3

Using different ranges of the possible values of a variable to invoke different functionality was very common when RAM and disk space for data and program code were scarce. Nowadays, you would use a function or an additional, accompanying value (boolean, or enumeration) to determine the action to take.

Current OS's suggest 1GiB of RAM to operate correctly, when 256KiB was high very few years ago. Cheap disk space has gone from hundreds of MiB to multiples of TiB in a matter of months. Not too long ago I wrote programs for 640KiB of RAM and 10MiB of disk, and you would probably hate them.

I think it would be good to cope with code like that if it's just a few years old (refactor it!), and denounce it as bad practice if it's recent.