82

When answers are posted within a question's initial grace period, it is possible that further edits could completely change the question (see "chameleon questions"), and render those first few answers incorrect.

Worse yet, nobody is notified that there was a change to the question, except those that are sitting on the page and notice the edit bar pop in. And of those, only those bothered enough to open it in a new window to compare the differences, will see how much was changed.

After the grace period is up, nobody will be able to consolidate why the first few answers actually answered a different question. And the answerers won't know unless someone comments or down-votes (and in the latter case, without context, they still might have trouble understanding, unless they memorized the question's initial state).

Jon Skeet stated it pretty well in this comment on a different feature request:

I'd like this to apply to the question, as well as answers. Sometimes I (or others) have added perfectly good answers, but then the question has been changed within the grace period, making answers look foolish without any indication of what's happened.

So this time, my proposal is quite simple:

Once an answer is posted to a question within the initial grace period, a new grace period should start, and the current state of the question should be saved as the first revision. Further edits within the next 5 minutes will be aggregated into a new, second revision.

This will only noticeably affect questions where answers are posted within the grace period and the author of the question makes edits within 5 minutes of the first answer. (Effectively, it would treat it as if a different user made an edit within the author's grace period.) All other scenarios will continue to behave as they do today.

This feature request comes close:

Add an indication that a post has been edited in the 5 minutes grace period

However, the answer suggests a simple change to the wording or a pencil icon (like we see with edited comments), without adding a revision that can be viewed and compared. I don't think this is enough - in some cases it's going to be important to know what changed, not just when.

  • 9
    What would make this request more actionable is if you could demonstrate some questions where this has happened. </snarkasm> – billinkc Jan 17 '14 at 16:13
  • 4
    @billinkc happened to me, too. But it was in my early days on Drupal Answers, so i can't find it using recently deleted list. I've seen a question, answered, then I have seen it again, flagged duplicate, one moderator closed it, then first one was edited, another moderator reopened second question, and I have started to accumulate downvotes as my answer was under wrong question... without any way to prove what happened. If any of the diamonds wants to search Drupal Answers, it will be there somewhere. – Mołot Jan 17 '14 at 16:30
  • @Mołot Bill is making a joke; there was a big hubbub about one of my other feature requests when a certain person asked for proof. The very nature of the problem makes it extremely difficult to prove. :-) – Aaron Bertrand Jan 17 '14 at 16:32
  • @AaronBertrand oh, OK. But there is a proof somewhere in my flags and questions' close and reopen history. Well, anyway, I just wanted to say it does happen, so I did. It still irritates me. – Mołot Jan 17 '14 at 16:34
  • 3
    @Mołot yes, agreed, I've seen it happen many times, but very rarely to me (I often write answers detailed enough that I'm not finished within the grace period). So coming up with examples might be problematic, but if anyone else has any, I'll be happy to incorporate them. – Aaron Bertrand Jan 17 '14 at 16:35
  • 12
    Need to think about this a bit more, but at first glance I rather like it. Avoids the UI clutter problem inherent in the previous request. – Shog9 Jan 17 '14 at 19:14
57

Jarrod whipped this up this past week, and the behavior should now be live. Edits will be rolled into the previous revision if the previous revision was created by the same author and none of the following conditions are present:

  • The previous revision was created 5 minutes or more in the past
  • A comment has been added to the post after the previous revision by anyone other than the editor.
  • An answer has been added to the post after the previous revision
  • The previous revision was a rollback
  • The new revision is a rollback
  • [as of April 3rd, 2017] The question was closed after the previous revision.

The way you phrased your request is a bit odd and does not quite match the implementation, but the intent is clear and I believe the changes satisfy the need.

  • 5
    Does condition 1, 2, 4 and 5 also apply to answers? – J. Steen Mar 23 '15 at 13:04
  • 11
    Everything applies to everything. Figure out how to post an answer to it, grace period ends. – Shog9 Mar 23 '15 at 14:43
  • 12
    he went to jarrod's! – swasheck Mar 23 '15 at 14:51
  • 1
    @Shog9 I was guessing it did - asking for clarification was mostly rhetorical since I thought it should be clarified in the answer itself... =D – J. Steen Mar 23 '15 at 21:53
  • 1
    Would you consider close votes and flags to be worth adding to the list of conditions? – mhlester Apr 15 '15 at 18:41
  • 1
    Shog, when a new answer is posted in a question, does this reset grace period of the other, earlier answer(s)? I am trying to figure whether this feature impacts issues discussed in Make first draft of a new answer part of the permanent revision history – gnat Apr 30 '15 at 22:55
  • No, @gnat. I still think that's crazy. – Shog9 Apr 30 '15 at 23:01
  • 4
    I think that fastest-gun is crazy – gnat Apr 30 '15 at 23:04
  • Unless "more pointless arguments" is your idea of sanity, this does nothing for that. – Shog9 May 1 '15 at 2:11
  • 1
    don't know if you noticed, but that very feature makes a solid progress towards what scares you (not me) so much. Any reader with 50 repz can now "freeze" what they see by commenting, that very much close if not stronger than saving first draft – gnat May 1 '15 at 11:00
  • @gnat: the 'reset grace period' feature doesn't stop changes being made. It just means that the changes are recorded when something around the question or answer changes, so that it is possible to go back in time and see the state of the question or answer when a comment was made (or when another answer was posted, or ...). – Jonathan Leffler Jun 6 '15 at 0:28
  • @JonathanLeffler agree, I meant to say something like a snapshot is frozen. Further editing is of course possible, it is just, as you say, changes are now "explicitly recorded". I already observed the way how new feature works, looks impressive. You see comments and revision history clearly reflects changes made to address the comment, it is not buried within grace period anymore – gnat Jun 6 '15 at 5:48
  • Should closing/opening the question also make new revisions? That can help clarify edits made in response to such an action. – Double AA Aug 16 '15 at 4:47
  • Shog, could you please take a look at this issue recently raised at MSO? They say that dupehammer close didn't break grace period, likely because "possible duplicate" comment was removed too quickly. If this was indeed the case, would it make sense to add question closure to list of events that reset grace period? – gnat Jan 6 '17 at 21:20
  • 1
    See the comment I left over on the discussion you linked, @gnat - the mechanical action is pretty common, but I see less evidence of abuse. – Shog9 Aug 11 '17 at 16:34

You must log in to answer this question.

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