Straw Poll: Return True or False?

[Joan M]

return HRESULT? ;P

[Maarten Verdouw]

Uhm, you could return false if something went wrong (No DB connection, index out of range e.g.). In that case a returncode or Exception would be more convenient. But that wasn't the question.

--------------- don't P A N I C

[Muammar]

Interesting question but I would always test the overall objective "that is the existence of the item in this one".

---

Smile: A curve that can set a lot of things straight! (\ /) (O.o) (><)

[Lost User]

This is a trick questions. None of the answers are fully acceptable. The question doesn't leave room for a third option of returning a status code which can take at least 3 values. Then, the answer would be to just return a separate, third value if the element to be deleted wasn't found! In my philosophy classes I was thought there are actually three states of truth: true, false, and undecided.;) The reason we're stuck with just true and false in programming seems to have something to do with the 0 and 1 values of independent bits.

[jlwarlow]

Christian Graus wrote:

If you return true for 'I found it and deleted it', and true for 'I couldn't find it, so I didn't have to delete it', what would return false ?

I found it but couldn't delete it. :confused:

Never argue with an imbecile; they bring you down to their level, and beat you with experience.

[Pete OHanlon]

public enum CollectionItemOperation
{
  ItemDeleted,
  ItemNotDeleted,
  ItemLocked,
  ItemNotFound
}
...
public CollectionItemOperation DeleteItemFromCollection(Hashtable ht, object item)
{
  CollectionItemOperation operationResult = CollectionItemOperation.ItemNotFound;
  if (ht.Contains(item))
  {
    ...
  }
  return operationResult;
}

There you go - no ambiguity.

Deja View - the feeling that you've seen this post before.

[Andrew Leeder]

In true K&R 'C' style you would return the number of items deleted, which would of course be zero. So in .NET terms that would be FALSE. ~A

[Guy Harwood]

I would return true. The outcome you were expecting is 'true', so return it! Does that make me a bad programmer? :-D

---Guy H (;-)---

[Mark II]

I know! I know! The *right* answer is... True! No, no, I mean false! Er... no. It depends. Um. Maybe neither? Or both? Or something. Yes, yes. Something. Or null? Right. Let's start with a bit of analysis. Requirements The question explicitly states that, the "function that is *meant to* delete an item". It does *not* say "is meant to delete an item *if* it exists.". The question does *not* say what the purpose of the deletion is. This could be: 1. To ensure that the item is not in the list. 2. To release the memory / disk space that holds the item. So, the non-deletion of an item must be assumed to be a failure of the function. Parameters Persumably there are the following cases: 1. The item exists and is deleted. 2. The item exists and cannot deleted. 3. The item does not exist. 4. The user has passed invalid parameters into the function. Case (1) is a success. Case (2) and (3) are a failure of the function. Case (4) is an error. This leaves 2 reasonable behaviours: A. Return nothing for (1) or throw an exception for (2),(3) and (4). B. Return TRUE for (1), FALSE for (2) and (3), and an exception for (4). Fowever, the question asks if we should throw TRUE or FALSE for (3), so (A) is not valid. Conclusion Answer: Return FALSE. And *DOCUMENT* the meaning of this result, ideally in XML coments so that they show up in intellisense. So, what do I win? My Blog: http://allwrong.wordpress.com[^]

[The Nightcoder]

Three methods: - DeleteItem: return void, throw one exception if item doesn't exist and another if the delete failed for some other reason. - ItemExists: return true/false. - TryDeleteItem: return true if the item exists and was deleted, false if it didn't. Throw an exception if the delete failed. This is consistent with framework behaviour, and gives the class consumer the option of picking whatever best suites the need. Note that although TryDeleteItem seems like a shorthand for an if, it might potentially be a performance optimization (if the collection is a database table, for instance). Then, DeleteItem would probably be a two-line wrapper for TryDeleteItem. Later, Peter

[Christian Graus]

OK, under what circumstances is that possible ? I have a linked list of any type, under what circumstances would you expect I'd be unable to remove an item ?

Christian Graus - Microsoft MVP - C++ Metal Musings - Rex and my new metal blog "I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )

[Christian Graus]

That's right, the question didn't indicate that the method did anything more complex. And, if it did, it's possible an enum would be required if a return value was needed to indicate what occured.

Christian Graus - Microsoft MVP - C++ Metal Musings - Rex and my new metal blog "I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )

[Christian Graus]

OK, so it's possible to think of a situation that's way more complex than the original question, and in this instance, it would make more sense for the return to be an enum, to return found it, didn't find it, or various possible other error conditions.

Christian Graus - Microsoft MVP - C++ Metal Musings - Rex and my new metal blog "I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )

[realJSOP]

I'm old and still writing C++ - I'd return integer that would indicate this way 0=success, 1=failed/invalid index, 2=failed/item doesn't exist, and I'd set up some nifty constants to match the possible return values so the code was more maintainable/readable.

"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt, 1997
-----
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001

[jlwarlow]

Christian Graus wrote:

OK, under what circumstances is that possible ? I have a linked list of any type, under what circumstances would you expect I'd be unable to remove an item ?

It would depend on the items in the list, if these items referenced a resource or database entry for example, and in this case the resource was unreachable or database was locked etc. Then you could fail to delete. Another example would be if the entry had a reference count that was higher than two meaning it was still in use. On reflection, throwing an exception would be a preferable to give a reason why the item could not be deleted. I think this has been mentioned in other replies in the thread.

Never argue with an imbecile; they bring you down to their level, and beat you with experience.

[Christian Graus]

An exception can be expensive, it all depends on how often it may occur. But yes, I said from the start that it depends on the complexity of hte method. Interesting discussion that Chris started.

Christian Graus - Microsoft MVP - C++ Metal Musings - Rex and my new metal blog "I am working on a project that will convert a FORTRAN code to corresponding C++ code.I am not aware of FORTRAN syntax" ( spotted in the C++/CLI forum )

[Aamir Butt]

I will return S_FALSE :)

"Some people believe football is a matter of life and death. I'm very disappointed with that attitude. I can assure you it is much, much more important than that. -- Bill Shankly"

[jlwarlow]

Christian Graus wrote:

An exception can be expensive, it all depends on how often it may occur. But yes, I said from the start that it depends on the complexity of hte method.

From my COM/DCOM days, you could return a HRESULT, S_OK for found and deleted, S_FALSE for not found assumed deleted, and failed codes for each of the reasons for failure ;)

Christian Graus wrote:

Interesting discussion that Chris started.

Indeed! :-D

Never argue with an imbecile; they bring you down to their level, and beat you with experience.

[DavidNohejl]

Depends on what do you mean by return value. Does it indicate if everything was ok (true) or something went wrong (false)? In that case, return TRUE if item to delete wasn't found. Or, does return value indicate if it deleted item or not? In that case, return FALSE if item to delete wasn't found.

---

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus "Real men use mspaint for writing code and notepad for designing graphics." - Anna-Jayne Metcalfe

[Bogdan Iosif]

The function should return FALSE. Choosing to return a boolean value from such a function implies the fact that I use exceptions to handle errors. Otherwise, it would be a bad design decision to return a simple boolean value. The function outcome is expected to be this: The value is certainly removed after a call has been made. The return value tells me a nuance of how the operation has been executed - TRUE if the value was actually removed, FALSE if there wasn't anything to remove - and I may or may not use this extra information in my code. However, the return value is not there for error handling. If anything doesn't go according to the expected operating mode - For example the value to be removed has a state that doesn't allow removal - I will throw an exception. If I can't use exceptions then the return value should be an integer with return types documented. For example, I can use 0 for success and anything else for various error codes. This is the model widely used in Win32 APIs. HTH, Bogdan P.S: I'm glad to see this kind of questions are being asked.