13

In other words, should we now look at the content of the answers when we are evaluating questions for possible duplication? Is it sufficient to say "Your answer can be found in this question over here, therefore your question is a duplicate?"

The new guidance is here:

Changes to "close as duplicate" (part deux)

Among other things, it describes the new close banner:

This question already has an answer here:
Link to the Duplicate Question

... which seems to suggest that, if some other question in the system answers the OP's question, then the OP's question can be closed as a duplicate on that basis (even if the question itself is not necessarily duplicated).

Example Here:

https://softwareengineering.stackexchange.com/q/225341/1204

Note the comments below the question, where I dispute the assertion made there that it is a duplicate question. The original question doesn't mention unit testing.

  • regarding the question in question, it smells an XY problem. Unit tests in isolation do not make a reasonable goal to achieve, their purpose is to support development of a new functionality. Eg, unit tests make little sense a project having no feature requests / bug fixes planned for implementation – gnat Jan 27 '14 at 21:49
10

Is it OK to close a question as a duplicate, even if the questions aren't the same?

Absolutely.

We do it ALL THE TIME on Server Fault when we send people to the Canonical "What I've been hacked! What do I do now?!" question, or the various Canonical Capacity Planning quesitons. There are thousands of "capacity planning" problems and infinite ways to be hacked, but the answer is always going to be the same basic steps...

A monkey of moderate intelligence should be able to read the linked question & answers, and make the logical conclusions necessary to apply that information to their environment (and if they can't they're probably the kind of person who wants a "gimme teh c0dez" spoonfeeding of exactly what keys to hit in what order to get their desired result).


Is this particular question a "duplicate"? (by either definition)

I don't know, but I'm inclined to agree with you that it isn't.

This question is talking about changing a process (how to integrate unit testing into the build/development process) - the proposed duplicate is about changing code (dealing with a mess). They're related, but not the same (you could possibly graft unit testing into the build process as a bolt-on).

I can definitely see how all the things discussed in the proposed duplicate apply, but I don't know that they necessarily add up to "an answer". I'd be more inclined to post an answer on this question (possibly linking to one or more answers on the proposed duplicate) than close it as a duplicate.

  • 1
    It also bears noting that this would be a Double-Duplicate (the target is already closed as a duplicate linking to two other questions (which are both centered on the "cleaning up code" aspect rather than the "modifying the build process" aspect) - how far down the link rabbit-hole do we want to send someone? – voretaq7 Jan 24 '14 at 20:27
  • 2
    I've been tempted to try to find the duplicate graphs on MSO at times. There are quite a few "double" duplicates and even "triple dups" of multiple levels. – user213963 Jan 24 '14 at 21:24
  • 2
    @MichaelT I only object when the graph contains cycles. Duplicate graphs should be acyclic! – voretaq7 Jan 24 '14 at 22:03
  • 2
    +1, can't agree more to both points of your answer. That's exactly what I was thinking when I read the comments of gnat and Robert Harvey. – Doc Brown Jan 24 '14 at 22:24
4

In addition to @voretaq7's answer, I would propose a methodology for evaluating if two questions are duplicate:

  1. If the two questions ask after the same problem, then they are definite duplicates
  2. Otherwise, if the topvoted answer(s) of one question provide an answer to the other question, then the two questions can be regarded as duplicates.

I propose to restrict the evaluation of answers to the topvoted answers, because very few people will look at all answers of a proposed duplicate to see if it really answers their question (or reviewers to determine if the vote as duplicate was correct).

If you think a lower-voted answer on another question answers a question, then you could always add a comment referring to the answer or write a new answer based on it.

  • 2
    With the possible exception of canonical questions (which are specifically designed for this purpose), I don't think you should ever look at the answers to evaluate question duplication. – Robert Harvey Jan 27 '14 at 21:51
  • 2
    @RobertHarvey: Why would canonical questions be different here? I would expect those to be prime examples of where you don't need to look at the answers to evaluate duplication. – Bart van Ingen Schenau Jan 28 '14 at 7:27

You must log in to answer this question.

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