17

I've seen this more than once. A question with a very weird dependency issue, got an answer to it's specific problem, and later on (sometimes months), someone votes to close it against the super general and unhelpful "How do I resolve unmet dependencies after adding a PPA?". This wastes reviewers time and close votes that should be used against other close reasons. If two questions whose answers can't be merged, is very likely they are not duplicates.

I know people know that I hate that Q with all my soul (along with others mega questions), and I will offer 5 questions about how to check dependencies issues in a way that any reader is capable to help himself instead of using carpet bombing.

  • 3
    AFAIK it is the OP's responsibility to prove(at the time of first post itself) that his/her question is not dup of How do I resolve unmet dependencies? – Registered User Mar 31 '14 at 14:26
  • 1
    @AdityaPatil if he accepted an answer that is not contained in the target dupe, what else he needs to prove? I'm talking more about old answered questions. – Braiam Mar 31 '14 at 14:27
  • 9
    @AdityaPatil Duplicates should be used when the question is the same, not the answer. – Oli Mar 31 '14 at 14:28
  • @Oli that too.. – Braiam Mar 31 '14 at 14:29
  • 2
    @Oli I never said the other way. I just mean that the OP should have researched(that's what we expect), found the question(it will pop up, I guarantee) and then tried the answers. If they don't work, he/she should post that this-that questions(and there answers) don't work for me.This way, no one would flag it as dup. – Registered User Mar 31 '14 at 14:31
  • @AdityaPatil well, and it happened again askubuntu.com/q/441537/169736 – Braiam Mar 31 '14 at 15:14
6

To expand further on what Oli said in the comments, I would argue that duplicates should only be used when both:

  • Any answer to question "A" would also be a valid answer to question "B"; AND
  • Any answer to question "B" would also be a valid answer to question "A".

If either of the above two conditions aren't met it's not a duplicate question.

In the case of a more specific question being nominated for merging with a more general question like in Braiam's example, that should not be considered a duplicate because answers to the more specific question would not all be valid answers to the more general question. The same would be true regardless of which one would be merged into the other one - that is, if the more general question is merged into the more specific question it still remains a problem.

You can certainly still refer to a more general question as part of an answer or comment on the other question.

Remember it's questions that you're merging, not answers.

The questions can be worded differently or have different details that would not affect the answers given, and still be duplicates if any conceivable answer to one would also answer the other.

Naturally, very specific questions are rarely going to be suitable candidates for merging because their specific details will be relatively unique in ways that would affect how people answer it. This is a good thing; we don't want to bleed the diversity out of questions by removing what makes them specific. There is a place for specific questions here just as there is a place for general questions. In theory people will be searching for content based on specific keywords related to their problem, so if there's a question that's a more specific match to their problem it should in theory come up highly ranked.

  • 1
    If the new question is just a more specific instance of the old question, and the new answer could reasonably be arrived at using the more general answer in the old question, then it's still a dup. – psusi Apr 3 '14 at 23:45
  • @psusi I think he said that in this part "The questions can be worded differently or have different details that would not affect the answers given, and still be duplicates if any conceivable answer to one would also answer the other." – Braiam Apr 4 '14 at 13:47
  • @Braiam, that seems to be a much stricter definition. If the details given in the question cause the answer to be slightly different than any answer in the other question, and, due to the different details, would not really be a good answer for the other question, it would fail this test. – psusi Apr 4 '14 at 14:04
  • @psusi check this blog.stackoverflow.com/2010/11/… – Braiam Apr 4 '14 at 14:18
  • 1
    Yes, this is a strict definition, and that is the point I was hoping to make. Any question with a significant level of specific detail will be unlikely to be a duplicate of anything, but that's good for the site: something should only be a duplicate if it's basically the same question. For example, "How do I stop Apache loading at boot" and "I've install Apache via apt-get install apache2 but it starts at boot all the time, how can I have it only start manually" are duplicates. "How do I install mod_security on Apache 2.4.9" is NOT a duplicate of "How do I resolve unmet dependencies". – thomasrutter Apr 5 '14 at 12:27
  • It's infuriating, actually, to see anything marked as a duplicate of "How do I resolve unmet dependencies". That question is so broad and generic as to be obnoxiously unhelpful: its top two answers are basically a novella-length comprehensive guide to the internals and configuration of APT. They belong in a wiki, not a question-and-answer. In terms of its long-windedness it's the antithesis of actually solving any specific problem with a specific package, and would only be of interest to someone who instead wants to simply "learn all there is to know about APT". – thomasrutter Apr 5 '14 at 12:35
  • If the same standards were applied to that question as are applied elsewhere, that question should have been closed as "too broad" a long time ago. – thomasrutter Apr 5 '14 at 12:36
  • @neon_overload I tried to close it, maybe you should ask a meta Q about it. – Braiam Apr 9 '14 at 0:32
  • If an answer is so long that it's not useful, click the "This answer is not useful" button (i.e. downvote) – Ben Voigt Apr 12 '14 at 21:04
  • 1
    @BenVoigt the top answer's currently at +158 and the question is at +168, I'm going to make no difference. Clearly I'm in the minority regarding how I feel about that question and its answers. To that I can only suggest that "huge wall of text" style answers tend to get a disproportionate number of upvotes just because of their length, however un-specific. I guess that's a different conversation though. – thomasrutter Apr 13 '14 at 7:19
  • @neon_overload don't feel so alone, I'm with you. – Braiam Apr 14 '14 at 3:10

You must log in to answer this question.

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