0
\$\begingroup\$

In this question I added a segment to the code scope only after describing the essential addition before the answerer in comments and getting his consent to make this addition in a reply comment.

I didn't delete any existing code or changed any existing code and this is something I would generally never do here as I fully understand why it goes against the nature of this site.

I was then, quite rightfully commented by @Mast that I should avoid this kind of editing but it is okay this time as the answerer applied this addition.

This brought me the idea to add to this site a button prevalent only in edit mode like:

The answer's author agreed to this edit in question

The button could be used only if code (scope) was expanded (existing code wasn't deleted or changed). The edit could be saved only after the button was checked.

Opinions?

\$\endgroup\$
  • \$\begingroup\$ What problem would this solve? Ordinarily, we disallow any code edit. What happens when there are multiple answers? \$\endgroup\$ – Mast Sep 11 at 9:29
  • \$\begingroup\$ This will give an ordinal way to prevent this specific kind of edits which I believe might be okay. Sorry for not thinking about a case of 2 or more answers; this is a real fuck up... \$\endgroup\$ – JohnDoea Sep 11 at 9:42
  • 1
    \$\begingroup\$ Considering the community has to judge these edits on a case-by-case basis to avoid a royal mess, it's slightly more complicated than that. I can imagine it's not a bad idea from your perspective, so I don't blame you for the idea, but I'll write an answer later today explaining why editing answered questions leads to much more complications than it seems. \$\endgroup\$ – Mast Sep 11 at 9:58
  • \$\begingroup\$ to prevent = to allow and regulate* sorry. \$\endgroup\$ – JohnDoea Sep 11 at 20:33
6
\$\begingroup\$

For the record, the full comment:

For the record, usually we're against updating the code in the question after answers have been received. It creates a mess if more answers come in, reviews of all different kinds of versions make it hard to follow what's going on. Please see what you may and may not do after receiving answers. However, since the author of the only answer so far is ok with it and the changes are minor, it's fine this time. Just, try not to do it again, ok? Thanks :-)

There's limited community consensus about when to and when not to rollback. The clear cases are easy, but there's a lot of not-so-clear cases.

In general, it's simply not accepted that code is modified after answers come in. Why? So every reviewer gets the same version of the code to review. Which is important, so if anyone finds your question after a year all reviews are about the same version of the code. It tends to get a mess otherwise.

There are however a couple of examples where allowing modification of a piece of code is "the lesser evil". Or where it's simply not worth the hassle to roll it back. I've been involved with a lot of rollbacks, and too many of them end in complaints, "recidivism" or other situations which require extra effort to resolve. Or when the question was actually off-topic, but OP and answerer teamed together to fix both the question and the answer. At which point a rollback would be pointless. So often enough cases are condoned, tolerated. That doesn't make them right.

The overall advice is not to make edits after answers come in. Adding a button for this very specific situation would just about legalise it under certain conditions and send the wrong message altogether. Would go against site policy.

You don't just post a question. You post a piece of code that does something, someone tells you it can be improved (and how) and you learned something. Important here is that other people learn something when reading both question and answer too, otherwise we might just as well destroy every question older than X weeks. We have to keep readability for those people in mind too, and sometimes that means keeping the mistakes intact and not allowing further edits when answers have arrived.

So, please no.

\$\endgroup\$
7
\$\begingroup\$

No. No custom button for "answerer approved edits".

I have multiple concerns. The primary one being that code-editing in (answered) questions is complex enough as it is, and the default case should be "Just don't". Adding a button is an indication that code editing is now "normal". Code editing is an exceptional thing, and anything that makes the impression that it is normal, or expected, should be prevented.

In addition, answerers are often new to the site too, and their perspective may not reflect the community, so their approval is not the only approval needed.... and what's to prevent the person from just checking the box anyway even if the answerer does not "approve".

Finally, technically, I suspect it would be hard to implement, and practically impossible to get the Stack Exchange staff to do such exceptional coding in the UI for just Code Review (I can't see how this functionality would benefit other sites in Stack Exchange).

\$\endgroup\$
  • 1
    \$\begingroup\$ Thank you for the great answer too. \$\endgroup\$ – JohnDoea Sep 11 at 20:40
  • 1
    \$\begingroup\$ Agreed - don't normalize such edits. But do please write (in the rare occasions that this is the lesser evil) in the "what you edited" field to help guide those of us who look at edits. \$\endgroup\$ – Toby Speight Sep 17 at 10:55

You must log in to answer this question.

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