7

I recently asked a question about linebreaking in a bibliography and it got an answer that appeared good. However, when I tested it, it appears to only work with the specific example text I provided. How do I update the question to respond to this? Sorry if this is general knowledge: I've never asked a question where at least one of the answers didn't basically work before, or at least close enough to what I wanted to modify into the code I need.

So, when I made the question I stripped things down as far as I could while still getting what I thought was a useful test case, that is, one bibliography example that shows the problem behaviour with that specific page geometry.

I got an answer that expanded that to three test cases, all by manipulating the text inside the DOI. Looks great I think, and today I go to test it..and it doesn't work. Or rather, it does, some of the time, and messes up most of my bibliography the rest of the time. What happens is the answer breaks if there is too much or too little text preceding the DOI. It will add a whackload of white space, or shove the DOI to a newline despite there being lots of space on that line for it.

How do I respond to this?

  1. Edit the original question to add more test cases? (This would make it much less minimal, and I'm not sure about editing the question to respond to a response? Is that OK? It would also look strange if someone else then responded with a totally different method, wouldn't it?)
  2. Respond in the comments (Hard, can't give an image or sample code. I've left a comment about how easily it breaks)
  3. Make a totally new questions (This feels wasteful, given the QUESTION is the same, I just only provided an example of the problem behaviour in the MWE, not a set of test cases)
2
  • 5
    My inclination would be to leave a comment to the answer saying that it works in only the cited example, and that you are posting a new question. In the new question, refer to the original question, say why the first answer doesn't work, and give a more extensive example that includes the original instance as well as others that don't work, each in a different way. Then go back and give a link to the new question in the incomplete answer. "Minimum" needs to show the complete problem, and that may not be "minuscule".
    – barbara beeton
    Oct 17, 2020 at 2:54
  • 2
    there are two answers there and neither give changes specific to that doi they both give general changes, \raggedright (which I'm surprised doesn't work always) or a usual penalty balancing trick, which again is general. The actual penalty may need adjusting to handle more tricky cases but I think it is misleading to say the people answering only addressed the specific example not the general issue
    – David Carlisle
    Oct 17, 2020 at 10:23

2 Answers 2

Reset to default
6

The consensus on main meta seems to be that you can edit your question, unless it invalidates existing answers. See Doesn't the asker have the right to modify his question? and Edited question makes answers incorrect for example. If it does invalidate existing answers then you should ask a new question.

The reason for this is that the invalidated answer may attract undeserved downvotes (because it does not appear to answer the question as stated), or in general it may confuse future visitors that have the same problem as in the new version of the question and find an answer that does not apply.

Of course here on TeX.SE we are not very inclined to downvote at all, and a comment or edit to either the question or the answer explaining that the answer applies to a previous version of the question will most likely prevent any downvotes or confusion. It also allows the answerer(s) to modify their answer to the new situation.

So, 'officially' you should ask a new question, but in this case both options (edit the current question or ask a new one) are fine I think. So it's up to you :)

1
  • 3
    While I agree with this in general, I don't think it should apply to the case outlined by the OP. I'd say, if just the editing of the MWE makes an answer "wrong", without changing anything else about the question, then that's probably just a clarification and should be fine. Even if answers already exist. I have often seen poorly crafted MWEs that prompted users to ask for clarification in the comments. When other users give a quick answer that then turns out not to be applicable to broader cases (including the questioner's), that should not necessitate a new question.
    – schtandard
    Oct 17, 2020 at 19:17
10

A minimal example is intended to help debug an issue by making it clear where the problem lies. So if a 'MWE' is too small to show the issue in real use, I feel it's appropriate to go back and edit. I'd probably make it clear that the original MWE was solved by the answer but that this didn't address your real issue, then provide an updated MWE.

This is not the same as 'changing the question', rather it likely shows that there is a subtly in the issue that the MWE you first created does not capture. An MWE is there is enhance a question, it's not the question in and of itself.

You must log in to answer this question.

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