7

This question was identified as a duplicate today and deleted an hour later.

There are three reasons why I believe deleting this was wrong:

  • As a general rule, deleting duplicates is a bad idea, as they serve as signposts (cf. this old discussion)
  • This question isn't really a case of exact duplicate anyway; I am not even sure if it was correct to close the question in the first place
  • There was a discussion about the correctness of one answer still going on at the time when the question was closed. Deleting in this situation can't be appropriate.

Do you agree that deleting posts in this manner and under circumstances like these is clearly wrong?

  • 2
    Goes back to my comment here: meta.stackexchange.com/questions/168563/… (sigh...) – Mysticial Mar 5 '13 at 9:02
  • 1
    That title is pretty localized and doesn't exactly serve as a very good signpost. Who would be searching for anything remotely similar to that? – animuson Mar 5 '13 at 9:02
  • IMO, no question should be deleted until the OP delete it themselves. – hjpotter92 Mar 5 '13 at 9:03
  • 8
    @BackinaFlash We would have hundreds-of-thousands of junk posts doing nothing but making our Google ranking suck, if that were how it worked. – Andrew Barber Mar 5 '13 at 9:04
  • @animuson It (the title and the body) serves as a signpost because it contains wording that is different from the other questions. Unfortunately, there are many ways to ask about this problem, and, even worse, not all of these situations are really completely the same (as the discussion on the main answer shows). – jogojapan Mar 5 '13 at 9:07
  • @AndrewBarber Duplicates should still be closed and marked as such. It's up to Google to provide search results that take this into account. – jogojapan Mar 5 '13 at 9:09
  • 1
    @jogojapan: A signpost that no one will ever see is not a very good signpost. That's like putting a stop sign behind a tree. – animuson Mar 5 '13 at 9:10
  • Not sure how that question would really be helpful in the future, especially with such a vague title. Perhaps it would have been beneficial to have edited it. The discussion seemed to be about semantics. Also, for long discussions, why not just go to chat so you can actually get a conversation in about the issue? – Travis J Mar 5 '13 at 9:41
  • 1
    @TravisJ Yes, both is true. We can move the discussion to a separate chat, and the question itself certainly isn't a very good one. But does the possibility of moving to a separate chat, and the fact there are a few issues with the question (which can be resolved by editing it) justify deleting the question? – jogojapan Mar 5 '13 at 10:04
  • And now, here's today's version ++++iValue Vs. iValue++++ – Bo Persson Mar 6 '13 at 10:47
  • @BoPersson Beautiful :) And it's got a reopen vote already... (not mine I promise) – jogojapan Mar 6 '13 at 10:59
2

There are a few issues here, on all sides.

Issues with the Question

  • The title screams 'too localized'. How would someone searching on the internet react if they tried to search for the problem this user had? All they see is text that says, "Why does the following code display 8 and not 9?" This makes the question virtually un-indexable.
  • The question itself is indeed too localized (as written). The intent itself may not be too localized (what is the following expression an example of undefined behavior in C++?) but the question as written is.
  • It's a duplicate, but as written (see above) it's not a good duplicate. Is someone really going to search for "Why does the following code display 8 and not 9?"

For those reasons, the question should not be opened, and really, adds no value to the site unless it was heavily edited.

If it adds no value to the site, why is it here?

Issues with the deletion

The C++ crowd (much like other 'large tag' crowds) has to deal with a lot of questions that are exactly the same. Sometimes it feels like the only way to combat these questions is to close and delete them as quickly as possible. I understand that. I may not agree with it (insofar as I believe there's a better way), but I understand it.

The question probably should probably have not been deleted because the OP can't see the question after its deletion unless they have a link to the question itself.

Both sides can be happy if the user can see their own deleted questions; then it wouldn't matter if it was deleted off the site, the user can still find it easily.

I agree that it should be deleted, but I think that it shouldn't have been deleted so quickly, not unless the above feature request was implemented. Once that's implemented, delete away.

  • Thanks for this, very useful, and very interesting. One quick remark about the title -- although I agree of course that it clearly needed editing: Finding a relevant question, e.g. by searching Google, is not only about matching your intent with the titles of existing posts. Matches with the body of the questions, as well as with existing answers (not to mention links to the question, if there are any) are also important. – jogojapan Mar 5 '13 at 14:19
  • @jogojapan See my second bullet point from the top. I agree. Unfortunately the body of the question does no better in that regard. Titles though are Really Important to Google, so it's good to get those right. – George Stocker Mar 5 '13 at 14:37
  • 2
    I agree that deleting questions this early is not a good idea, however I also argue this is a special case and something that is not done otherwise. Some time ago I spent several days worth of delete votes to bring the number of duplicates down from about 100 to 64. That number is more than sufficient to cover all variations of subject lines and ways of writing the code, including using i++, x++, y++, j++, a++, u++, and v++, and then some. And even though some new questions have been deleted as quickly as the one today, we are now back at 78 duplicates anyway and a bit desperate. – Bo Persson Mar 5 '13 at 15:39
3

This is a specific case for the C and C++ tags, where this question has been asked in more that 100 very slight variations already. New questions appear several times a week, sometimes more often than once a day.

Some of us has just gotten extremely tired of this and have started to delete them as fast as possible. If nothing else, it saves the OP from 20 downvotes from other frustrated users.

As a general rule, deleting duplicates is a bad idea, as they serve as signposts

Despite deleting a large number of these duplicates, there are still 78 closed questions linked to the master question. A couple dozen other questions are linked to one question about undefined behavior and sequence points.

We just don't need any more of these questions!

This question isn't really a case of exact duplicate anyway; I am not even sure if it was correct to close the question in the first place.

We already have the other 78 variations of the same problem. This is not code you would write in a real program, and it just doesn't work. Both language standards explicitly say so.

There was a discussion about the correctness of one answer still going on at the time when the question was closed. Deleting in this situation can't be appropriate.

I didn't really see the discussion in the comments. However, the code just doesn't work (because the C and C++ standards both explicitly tell us that it is not defined) and the exact details of why it doesn't work this time is not all that interesting. And is has been discussed a hundred times before.

There is absolutely nothing useful in code like ( a + func(a)++ ), u = u++ + ++u;, foo(i++, i++);, printf("%d %d %d",i,++i,i--);, printf("%d%d%d%d%d%d",i++,i--,++i,--i,i);, printf("\n %d %d %d ", a, a++,++a);, printf("%d %d %d %d %d",a--,a,a=20,a++,a=39);, etc, etc, that we already have.

Saving yet another copy doesn't make it any better.

  • I am not trying to defend these questions because I think they exhibit useful code. I am not really trying to defend these questions at all. I am talking about deleting something within 1 hour that a 100k++ user (not me!) found interesting enough to write a detailed answer for, which unfortunately seemed not entirely correct and therefore triggered a discussion. – jogojapan Mar 5 '13 at 14:30
  • Anyway -- although this meta question here is about a rather specific situation, I definitely agree that the problems posed by these extremely frequently repeated questions are enormous. The sequence point/UB is one example, unexpected floating point results is another one. I am not convinced that the whole duplicate closing/deleting mechanism is the right way of dealing with this, not least because these things are never exact duplicates, so the close reason given when something's closed as duplicate isn't really accurate here. – jogojapan Mar 5 '13 at 14:44
  • 1
    That 100k user finds everything interesting enough to write an answer for. That's how he has earned his 100k. :-) – Bo Persson Mar 5 '13 at 15:42

You must log in to answer this question.

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