When is "update your earlier question" no longer appropriate?

Question

I find it hard to believe that this hasn't been asked before, but I wasn't able to find a dupe.

I've seen several new users go through these steps:

1. post a code sample, without any other info, that contains multiple bugs
2. get a fix for one of the bugs
3. recompile and get a different error
4. post a new question with nearly the same code but a different error

Usually, I and/or someone else will gently poke those users to let them know that they can edit their earlier questions instead of making a new post.

A certain new user made three such posts yesterday (one, two — 10k only, three) and just posted again about the same issue. I was about to post another comment asking him to edit an earlier post, but I noticed that his code today was somewhat different than it was yesterday.

Based on that code change, is this a legit separate question, even though it deals with the same topic discussed yesterday? One of the earlier answers even mentions code 450 vs. code 530. And, more generally, when is it appropriate to post a new question about a topic you've recently asked SO/SF/SU about, instead of editing the earlier post?

Answer 1

If it is a different question that doesn't rely on the information from the first question, then you should ask it as a new question. For example, I once asked two separate questions about the same section of an application I am working on. However, the topics they cover were completely different and aside from the basic background on what that section of the app did, there was no need to share any information between them.

But if a significant part of your original question is key to the following question, then it really makes better sense to stay in the same question.

Answer 2

I'm not sure that extended debugging sessions are the right kind of question to be asking on SO. I think they'll have little if any value to future users of the site. If I don't start out with the same piece of buggy code, then I'm not likely to have the same set of errors.

Having said that, I think it calls for a new question if it deals with a new error. Once you fix the first problem then the answers to the original question still apply to that problem. It doesn't seem right to change the question to make those answers invalid. It also seems like you won't get any new answers if you add new information about a new bug you've uncovered, because your question will already have multiple answers.

Answer 3

It's a hard line to draw, and usually needs to be determined on a case-by-case basis. It really depends on how different the new situation is. The same project can have many different questions.