4

Possible Duplicate:
“Close as duplicate” - what if only the answer is a duplicate?

I have just asked a question on SO: Why does casting a function to a function type that is identical except for return type fail? . One user commented that it might be an exact duplicate of another question ( Is the return type part of the function signature? ), and presumably voted to close it on this basis, as well.

I pointed out in a following comment that my question was distinctly different, and the response in a following comment was:

If you read the answer I referred to you'll see that it explicitly answers your question!

It seems odd to me that a question would be marked as an exact duplicate if the answer to that question is contained within the answer to a different question, even if the questions themselves are different. When I asked about this, a different commenter stated:

It’s a duplicate if the answers to the other question would equally apply here.

Again, it seems strange to me that the category exact duplicate should apply even if the question is different. For example, how should I (or anyone in the future) know to look at a different question for an answer to their question?

It seems to me that if a different question has an applicable answer, then the question itself should not be considered an exact duplicate; but rather, a good answer to the question would be either to link to the answer in the second question, to repeat or copy/past the relevant information from the answer to the second question, or both.

I looked at the StackOverflow FAQ for guidance. It says:

exact duplicate

This question covers exactly the same content as earlier questions on this topic; its answers may be merged with another identical question.

This strikes me as possibly ambiguous, but fairly clearly indicates that it is the content of the question, not the content of the answers, that distinguishes exactly duplicated questions.

I would like to gain clarity on the possibly ambiguous definition of "exact duplicate". Specifically, does exact duplicate mean that the content of the question itself is an exact duplicate of the content of another question (regardless of answers to the questions), or is it possible for a question to be an exact duplicate if the answer to another question contains the answer to the original question?

If, in fact, a question is considered an exact duplicate if the answer to it is contained in the answer to a different question, then I would like to understand the rationale behind this decision, given that people browsing or googling in the future will not have good reason to look at a given question if it is not the same question they have.

In my case, I did see the linked question before I asked mine, but I did not look at it because I already knew the answer - so this is a case in point. In this case, the answer to my question is (partially) contained in the answer to the linked question only because the person who composed that answer went above and beyond the call of duty to compose a thorough wiki-style essay that not only answered the specific question, but also touched on other areas as well. That seems like a good reason to link to that answer in the answer to my question - not to mark my question as an exact duplicate.

6
  • 2
    It makes more sense if you don't take the "possible duplicate" text too literally but more in the sense of: "here's your answer!". I believe there's no point in duplicating information, even if there's a different question to begin with, the path to the answer is clear.
    – slhck
    Commented Dec 3, 2012 at 16:03
  • 1
    @slhck I think most people search for questions that are the same as theirs. How should a person know to look at a different question for the answer to a question they have? Wouldn't the correct approach be to provide an answer that links to the answer in the other question? That way, people in the future will be drawn to the answer of their (distinct) question. Commented Dec 3, 2012 at 16:04
  • 1
    They can't know, but if we provide them with the link to it, I don't see the problem, even it the question is not quite the same.
    – slhck
    Commented Dec 3, 2012 at 16:07
  • @slhck That is my suggestion - provide an answer with a link to the answer in the other question. If the question is closed as exact duplicate, it implies that the question is an exact duplicate. It's not, in my case. Commented Dec 3, 2012 at 16:08
  • I meant a link as in the "duplicate" link. But if you need to give more context than just posting the link, a real answer paraphrasing the other would be more appropriate of course, you're right.
    – slhck
    Commented Dec 3, 2012 at 16:11
  • Though this question has been closed as an exact duplicate, it is not. This question asks whether a question is an exact duplicate if the answer is contained in (the answer to) another question - not if the answer is the same as the answer to another question. In the case I linked, the answer was only contained in the answer to another question because that answer was a wiki-style essay that contained extra information, and it was the extra information that answered my question. Yet someone still wished to close my question as a duplicate. That is my question here - not a duplicate. Commented Dec 4, 2012 at 0:25

2 Answers 2

3

The other user is not correct; an "exact duplicate" is a mechanism of stating that you are asking the same question. (Although "exact duplicate" ends up being "pretty close duplicate" in practice, very minor differences not germane to the answer are not considered.) If there happens to be an entirely different question that happens to answer your question the appropriate resolution would be to answer the question, provide a link the the existing answer, attribute it, and summarize/quote the relevant section of the other answer. (That last part is particularly relevant in this case since the other answer is much more broad, and only a paragraph or two of it is relevant to your question.)

Now, on a few occasions, particularly on meta, the community is a bit more loose with this. There are a few types of questions that end up being asked a lot on meta asking for information already found in "faq" posts, or other posts written to be canonical answers to these common questions. While they tend to not be "exact duplicates" all of the information is there, and many people just don't bother to spend a lot of time and effort answering questions already detailed in the FAQ. This is one of those times where "SO doesn't always follow it's own rules" and really shouldn't be taken as an example to be followed elsewhere on the SE network.

3

"Does closing as a dupe require that the questions are describing the same problem, or that both questions have the same right answer(s)?"

Both.

There are two key requirements that must BOTH be met for what I consider to be a legitimate dupe:

Another question is essentially asking the same thing.

That means that they're both highlighting the same problem or need, which is not the same as them both having the same answer. Consider the following two (silly and made up) questions:

What's the best way to seal a window in advance of anticipated torrential rain to prevent leakage?

and

What's the best way to seal a window to block the smell of chocolate chip cookies with minimal work and hardware?

The answer to both questions may well be, "use duct tape," but they're not really the same question. If they both were proxies for, "what's the best way to seal a window", you'd want to edit one to that, but they really aren't. Because depending what you're sealing against, the answers may be different, even though they don't happen to be different here.

That same other question's answers solve the new asker's problem.

If the old question's answers don't address the problem the new asker has, then one of two things has happened:

  1. The original question got good answers, but the needs in the two questions are different enough that thier answers differ. This actually means that they're different questions, in a way that the answer highlights better than the question did. In the rare case that the question and/or answers can be made generic enough that they'll apply to BOTH needs without losing the necessary detail for implementation and you can make the canonical question/solutions, go for it, but that's rare.
  2. The original question has no good answers. Possibly because it wasn't written well, or it simply failed to attract sufficient attention when it was written. This case is trickier, but closing the new one seems wrong. I'd personally probably wait in this case and see what happens with the new one. If the new one gets good answers, you can actually close the older, unanswered one as the dupe. (I can hear you getting ready to ask if it's fair. I'm not sure, but it's a darn good way to improve the knowledge base on the internet and help more people.)

Not the answer you're looking for? Browse other questions tagged .