5

I was just about to answer this question when it was closed. I'm cool with it being closed, but I think we should have some unified stance regarding lumping together questions by pattern.

A lot of these problems asked on dba.se have a similar work around for a specific rdbms or follow a pattern. In some cases, we even tag the pattern (such as ). Should we also not be doing that in the closed question? We closed it because it follows the pattern of the "loose-indexscan" and we already had a question that answered to that pattern. But, we don't close all of the questions that match Gaps-and-Islands. I'm wondering if we should either,

  • Create a policy that we tag the pattern, for better indexing. And we permit duplicate patterns but answer to the specific work-case.
  • Dupe all patterns, and vastly reduce the question size of DBA.se.

Thing I was planning to do,

  1. Show sample data.
  2. Show his current method.
  3. Show how the loose-indexscan workaround works.
  4. Show the new method.

All of that is specific to his use case.

  • Just an explanation why I closed it as duplicate: the 2 questions have almost identical queries (select "labelDate" from pages group by "labelDate"; and SELECT DISTINCT record_type FROM records;) and identical plans (if table/column names are replaced). There are several variations of queries that can benefit from the same pattern of solution (loose indexscan with recursive cte) but they are more or less similar (with of course some ore complicated problems with joins or multiple levels of grouping). – ypercubeᵀᴹ Jul 28 '17 at 18:52
  • The "gaps-and-islands" category is even more complex, with many related but different types of problems (find gaps, find islands, find closest island, concatenate islands, etc) with often many different queries/solutions. – ypercubeᵀᴹ Jul 28 '17 at 18:54
6

Closing questions of the same pattern... (wiki)

If we only ever closed exact duplicates, there would be very few of them.

Closing instances of the same pattern against the canonical Q & A is beneficial; we should continue to do that. The question author gets pointed to a great resource that should solve their problem, and our ability to answer search engine queries is enhanced by having a different phrasing of the question. Closure as duplicate without answers results in an immediate redirect for anonymous users.

...or answering them for the use case (forum)

We already have a system for this.

The first vote to close as a duplicate will alert the question author to the existence of the potential existing answer. The author has a binding vote to close as a duplicate if it solves their problem:

UI

If the question is different, it is up to the question author to make it clear in the question why that is.

Meanwhile, other users can vote to close (via the review queue as well as directly) or answer the specific question as they see fit. Either or both of these actions is acceptable, in my view.

Where the question is immediately marked as a duplicate by a gold tag badge holder...

Mjölnir

...the message shown to the question author is different, but the process is much the same:

The first question edit after closure will trigger an automatic reopen review, so we can assess whether the closure as a duplicate was correct or not. Again, the onus is on the asker to make the case.

If all else fails, the question author can always ask a new question, noting the existence of the related Q & A, but making it clear which specific aspect they still need help with.


Often, there is no way to know (from the question alone) whether the asker is simply unaware of the pattern and general solution, or has tried it and failed to apply it in a particular instance.

We are not psychic, so there has to be some indication in the question that a specific solution is required. We're reasonable people here. I'm sure we would co-operate to reopen the question briefly to allow such an answer to be posted in edge case scenarios.

This is all easier now that gold badge holders can dupe close and reopen with a single action.


There is always a certain tension between the forum and wiki aspects of the SE model. All possible outcomes noted above seem OK to me, so I don't think there's a need for a rigid policy here, beyond current practice.

Related FAQ: How should duplicate questions be handled?

4

Closing as duplicate should serve two purposes:

  1. The duplicate should link to a question with best-in-class answers.

  2. The duplicate is often worded differently and as such allows a wider variety of searchers to find a good solution.

If a question is blatantly a duplicate of another question, and the older question has excellent answers, it should be closed as dupe. If the new question has some wrinkle that makes it unique or is sufficiently different that a new, great answer would be helpful, I try to provide an answer. Sometimes a question gets closed while I'm in the process of writing an answer - in those cases I typically add a comment saying that I'd like to answer and why I think its not a dupe and should be re-opened.

  • I don't think I have anything special to add in the generic, nor would I argue that it's a wrinkle. It's just I was going to show up a test, sample code, and a benchmark for his use case. I think that would provide extra value to him. That's likely true of all question we close for following a pattern. – Evan Carroll Jul 28 '17 at 18:10
  • 3
    Close as Duplicate is indeed a bit problematic since every question is always going to have some difference. Closing as duplicate offers a quick way for us to help the OP without having to re-write the answer with slight modifications every time. – Max Vernon Jul 28 '17 at 18:25
2

The tagging (side-) question

is useful because it groups related questions.

I'm not certain is quite the same, since it might be more about a potential solution than the question itself.

We have established on meta before that answer-specific tags should not be used (even when the solution is obviously the only one applicable).

That said, I'm not certain it is so clear cut, so you are welcome to create such a tag to see if it takes off. It's not as if our tagging system is so perfect that it couldn't tolerate such an anomaly.

Perhaps you can think of a better tag name that describes the type of question more.

You must log in to answer this question.

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