C/C++: Naming conventions for arrays?

Question

What is a good naming convention for arrays? I am working on a code base with few thousand line of codes and there is no consistent naming convention for arrays. Few ppl name them by appending List at the end of the name like *message_list*, which I really dont like as it wrongly suggests that this is a list(linked list), and few people name them by appending 's' at the end like messages which is better than previous approach but sometimes fails.

I want to know if there are any naming conventions out there for array type variables?

Solution 1

Name them for what they hold. An array of messages should be called messages. An array of file names should be called filenames. Capitalize to suit your whims and prejudices.

There are those who argue for the singular, suggesting that message[i] is singular and better than messages[i]. There is some justice in that, but the counter-argument is that the variable names an aggregate, so the plural is better.

I'd not use a suffix such as _list for an array.

Solution 2

A naming convention we had at a previous job I worked at was variables started with some letter to indicate their scope (m_ for member variables, a_ for parameters, etc). Then after the underscore was another letter or two indicating the type (_p for pointers, _dw for ints (dword), _s for strings, etc) then the variable name. You could just adopt something similar to this, all that really matters is your code is readable to your convention, so you could start all of your arrays with something like a_VariableName to say that since it starts with an a_ it is an array

Solution 3

I rarely use "primitive" containers directly. Rather, most often, the array (or vector) is a member in a "real" class that is a "first-class-citizen" that logically identifies some logical aggregate. Thus, the name of the array is an obvious data-member name for the class:

class MyCritters
{
private:
  std::vector<Critter> critter_set_;
  //...or...
  Critter[10] critter_set_;
};

This has the added advantage that I have "intercept points" to perform logical set operations, ensure non-NULL, ensure uniqueness, add type-specific iteration algorithms, etc.

However, if I were to simply use an array directly, I typically append "set" to name. On my system, I have common interface functions for arrays, vectors, linked lists, etc., and those are all (ordered) sets. Thus, "set" will always be correct, and I can simply change the definition (instantiation), and all the other accessors still work correctly.

Comments

[Dan F]

Honestly, I got so used to it that I find code without it a little hard to read

[Cody Gray]

dwords are, of course, not integers.

[Dan F]

an int is a dword though...in most compilers

[Cody Gray]

Not in the Windows world. DWORD is always an unsigned int.

[Dan F]

Yeah, I recall now that we did use it for unsigned ints, but we also did most of our programming in PS3

[Thomas Matthews]

+1: "Capitalize to suite your whims and prejudices."

[Jonathan Leffler]

A mathematical set has no duplicates; arrays in general can have duplicates. I'd not use _set as a suffix unless the array was constrained to be a mathematical set without duplicate entries.

[charley]

@Jonathan, somewhat agree, but "set" is a standardized interface where the implementation may be "vector" or "linked-list". Change type, and recompile, interface does not change. By our convention, MyCritters is known to be a unique collection of Critter instances, while we typically also have MyCritterSet (as the data-member within MyCritters) which is not enforced to be a unique collection, may contain NULL, and may actually be implemented as a vector or linked-list (as an implementation detail).

[crayzeewulf]

Stroustrup's opinion on Hungarian Notation.