tags:

views:

899

answers:

13

This is re-posted from something I posted on the DDD Yahoo! group.

All things being equal, do you write phone.dial(phoneNumber) or phoneNumber.dialOn(phone)? Keep in mind possible future requirements (account numbers in addition to phone numbers, calculators in addition to phones).

The choice tends to illustrate how the idioms of Information Expert, Single Responsibility Principle, and Tell Don't Ask are at odds with each other.

phoneNumber.dialOn(phone) favors Information Expert and Tell Don't Ask, while phone.dial(phoneNumber) favors Single Responsibility Principle.

If you are familiar with Ken Pugh's work in Prefactoring, this is the Spreadsheet Conundrum; do you add rows or columns?

+12  A: 

phone.dial(), because it's the phone that does the dialing.

Actor.Verb( inputs ) -> outputs.

Jason Cohen
+1  A: 

If your writing OO then you start with the basic object, which is not the number, the number is going INTO the phone, so phone.dial() that way you can also phone.answer() phone.disconnect() phone.powerOFF, ect.

Another way to look at it is does the phone dial the number or does the number dial the phone?

Unkwntech
+1  A: 

Clearly, phone.Dial(number)

David Thibault
+7  A: 

Meh - User.Dial(number). The phone is meaningless in the given context. SOL (speak out loud) is a nice way to think this through (idioms and principles aside):

Phones have a dial. They can't dial themselves. Phone numbers are digits. Users dial PhoneNumbers on a Phone Dial.

Rob Conery
I don't think we are talking rotary phones. Any cell phone with a built in phone book can dial itself. User.Call(contact) -> Phone.Dial(number)
spoon16
A: 

Not to be the negative one here, but these kinds of questions are very academic. It completely depends on the application. I can think of very good reasons for doing it either way, and I've seen too many good programmers get bogged down in this kind of moot design details.

Lucas Oman
A: 

I'm not sure how that relates to the spreadsheet conundrum. Do you expect, in the future, to use phones to dial account numbers? To use phone numbers on calculators? Your example of "future requirements preparedness" is not very good...

Plus, you use the verb "dial". Sure, I could imagine "dialing" an account number on a phone. (It's a big stretch, though.) But if this phone number is to be used on a calculator, would you call the action "dialing"? If the name of the function changes depending on the type of parameter it gets passed, you have a design error.

In a typical OO design, objects get sent messages carrying data, not the other way around.

consumerwhore
I didn't realize there were comments for individual answers; please see my first comment I left on the original question. Thanks.
moffdub
A: 

phone.dial() +1.

What is the variant state or behavior of a PhoneNumber? The only thing that comes to mind are "dialing rules" (dial country code if outside, dial "9" to get outside line, etc.). That context seems well suited to the Phone.

If your object model doesn't require variance -- a number is just a sequence of digits, "dial" is just foreach(digit in phonenumber) { press(digit); } I'm with Rob Conery: meh.

Larry OBrien
+1  A: 

Clearly the PhoneUserInterface interface, which you can get an implementation of from the PhoneUserFactory.CreatePhoneUser() method, has a method dial(Phone, Number) that you can use to dial the phone.

EDIT: Answering the comment. Neither. The phone should have a buttonPressed() or something like that. The user enters the digits/characters of the phone number via that interface.

Jason Dagit
Right, and that code calls which one: phone.dial(phoneNumber) or phoneNumber.dialOn(phone)?
moffdub
+1  A: 

Neither. The User dials a Phone Number on a Phone.

Jeff Hubbard
Aside from smelling procedural, inside of that method, the question still stands: which do you call: phone.dial(phoneNumber) or phoneNumber.dialOn(phone)?
moffdub
Something more like: "for each(number in phoneNumber) phone.dialNumber(number)". A Phone has no reason to understand what a PhoneNumber is (object-orientedness aside, think about things like required vs. non-required area codes, international calls, etc). Dratted formatting.
Jeff Hubbard
+2  A: 

the question assumes the context of the answer, and thus creates a false dilemma

the 'spreadsheet conundrum' is a false dichotomy in this example: rows and columns are the presentation layer, not necessarily the data layer. The comments below tell me i misunderstood the analogy, but i don't think so - saying 'should this be a row or a column, which one is more likely to change' is forcing an unnecessary choice on the problem space - they are both equally likely to change. And in this specific example, this leads to choosing the wrong [yes wrong] paradigm for the solution. Dialing a phone is how old mechanical devices initiated a connection to another old mechanical device; this is hardly an apt analogy for modern telephony. And assuming that there is a 'user' to initiate the call simply moves the problem - although it moves it in the correct direction, i.e. away from the rotary-phone model ;-)

If you look at how the TAPI [sorry about the typo earlier, it's TAPI not ATAPI!] protocol works, there is a call controller - equivalent to the 'user' i suppose in some sense - that manages the connections between devices. One device does not call another, the call controller connects devices. So the example below is still essentially correct. It might be more correct to use a CallController object instead of a generic Connection, but the analogy should be clear enough as is.

In this example, a phone is a device with an address aka a 'phone number'. The 'dial' operator establishes a connection between the two devices. So the answer is:

Phone p1 = new Phone(phoneNumber1);
Phone p2 = new Phone(phoneNumber2);
Connection conn = new Connection(p1,p2);
conn.Open();
//...talk
conn.Close();

this will support multi-party calls as well, by overloading Connection to include a list of devices or other connections, e.g.

Connection confCall = new Connection(p1,p2,p3,p4,p5,p6);
confCall.Open();

Connection joinCall = new Connection(confCall,p7,p8,conn);
joinCall.Open();

look at the TAPI protocol for more examples

Steven A. Lowe
You misunderstood the analogy. It is not a literal spreadsheet, but rather you are lining up classes on the rows and columns, and seeing which is more likely to change in the future: rows (phone, register, calculator) or columns (phone number, account number, SSN).
moffdub
the question assumes the context of the answer, and thus creates a false dilemma; see edits above
Steven A. Lowe
I agree that this seems like the perfect fit for a service or controller. Usually when I'm stuck between two design choices and neither fits better than the other - I tend to think I'm not seeing the "right" choice yet.
qstarin
A: 

I wouldn't have phone number as a class at all as it does not have any behavior, it's just a data element.

Sijin
What about 109578390258792836? Is this a phone number? What's the country code? If USA, what's the area code? Do you need to dial a 9 first to reach the outside line?
moffdub
A: 

A: phone.dial(phone_number)

The PhoneNumber is dumb and is only a dataset. When the "dialling" happens, should the the PhoneNumber object know how to dial? There are many states to keep track of, like:

  • Is the phone already on another call? (if yes/no, what to do?)
  • What happens if the method of dialling changes? (global roaming, different carrier, etc.)
  • Also, what about scope? When a call is made, the phone number needs to be added to the list of recent outgoing calls.

If your PhoneNumber object needs to know all this, it's not DRY and your code will be less portable and more likely to break.

I would say that Steven A. Lowe has it down. This should be done by a Controller type object to handle the different states, etc. Keep your PhoneNumber object dumb and give the smarts to the middle-man who needs to worry about keeping the phone humming along.

Dan Harper - Leopard CRM
thank you very much, i feel validated!
Steven A. Lowe
+1  A: 

Choosing whether to give the column objects or the row objects the dial method doesn't change how the program will scale.

The dial method is just going to be itself a sequence of row and column methods. You have to ask what those methods depend on.

If the sequence of row methods doesn't depend on knowing exactly which column object is involved (but does depend on which particular row object is involved) and vice versa for the sequence of column methods, then the problem scales as m + n (m = num. rows, n = num. cols). When you create a new row it doesn't actually save you any work had the column method been assigned the 'dial' method. You still have to specify a unique sequence of row methods for use in 'dial' somewhere!

If, however, say the sequence of column methods inside 'dial' doesn't even depend on which column object is involved (they use one 'generic' sequence of column methods), then the problem just scales as m. It doesn't actually matter if you've assigned the 'dial' method to the column objects, the program still scales as m; essentially no work is required to make a new dial method when adding 1 more column object and you clearly have the option of abstracting all those dial methods themselves into one generic dial method.

dbs
I agree, it depends, and it definitely helps to abstract the types of the rows and the columns to help scalability.
moffdub