I noticed this anomaly after clicking the "load new answers" link on this question:

two accepted answers?

As far as I can tell, this was caused by the following sequence of events:

  1. I loaded the question with one accepted answer and left the tab open for a while.
  2. Meanwhile, a new answer was posted and the OP changed this to be the accepted answer.
  3. I returned to the tab, noticed the "load new answers" banner, and clicked it.
  4. The new answer got loaded, but now both were shown as accepted.

Of course, refreshing the page fixes the issue.

  • Can you reproduce it or did it only happen once?
    – Ben Brocka
    Apr 2 '12 at 3:28
  • @Rarity: Only once so far, but I suspect it should be reproducible by the steps above. It's kinda hard to test on my own, though, since people don't change the accepted answer all that often.
    – hammar
    Apr 2 '12 at 3:32
  • Oh, yeah if it's tied to changing the accepted answer it's probably extremely rare, I was wondering if it was just a live refresh problem on it's own
    – Ben Brocka
    Apr 2 '12 at 3:36
  • 9
    This isn't really a bug; it specifically says "load new answers", not "load new answers and update the current ones". Any changes made to the answers already on the page aren't going to be reflected Apr 2 '12 at 3:47
  • 12
    @MichaelMrozek - Despite the arguable nature of the steps (or the language chose for the live update of answers), would you have expected to see two answers marked as "the answer"? It most certainly is a bug regardless of how you dissect it. Perhaps not a logic or "by language" bug, but in the end the behavior differs from that which a typical user would expect, hence it is a bug.
    – M.Babcock
    Apr 2 '12 at 4:52
  • @hammar - Oh +1 BTW.
    – M.Babcock
    Apr 2 '12 at 4:53
  • @MichaelMrozek, I don't agree on your point - see my comment on Nick's answer.
    – Tomas
    Apr 2 '12 at 11:59

Accepted answers will now be updated as they happen. If a change is made to the accepted answer it will be distributed to all people who are looking at that post.

  • 1
    +1 - AAAAAA++++++ Would code with again!!11!!eleven!!1! Apr 3 '12 at 10:18

IMO this isn't worth fixing. In general, it adds a tremendous amount of complexity for what it solves. What if you have votes expanded? are editing one of the existing answers? have a mod or flag dialog open? Simply refreshing the existing answers isn't an easy thing to handle, it adds complexity to many things (and has to be accounted for in all new functionality we add).

However, if we integrated accepting an answer into the real-time vote updates or something along those lines, it'd eliminate the issue in other ways and give a net benefit. Something like that may be desirable, but refreshing existing answers isn't a path we'd necessarily want to go down.

  • 5
    Nobody expects that all answers will be refreshed after loading new answers. But the "accept mark" is kind of global information now loaded with new answers - and if the accept mark is not reset in the old answers, this newly loaded information itself is misleading. I feel this (and only this) shall be corrected. I guess it would be something like one jquery call in AJAX handler - if any of the newly loaded answers is accepted, just reset all previously accepted answers before updating the DOM.
    – Tomas
    Apr 2 '12 at 11:54
  • @Tomas - Sure, but that's drastic overkill. If you read the second half of this answer I was proposing a much more general solution that solves the problem (rather than addressing symptoms as they happen), which Geoff has now implemented. Apr 3 '12 at 10:17

You must log in to answer this question.

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