Translating array pointer access from C++ to Delphi
I agree that the question should not be deleted for minor problems like that (Should mostly working but a little bit broken code stay or be deleted?). But should the question be answered? I am afraid that reviewing broken code (no matter how simple would be to fix) is not healthy in the long term.
Would it be better for the reviewer to fix the minor problem directly in the question, before answering?
First of all, you must never fix a bug in the question just by editing the question. This is Code Review, not GitHub hack fest. (The only possible exception would be obvious trivial whitespace issues due to the author's inability to copy-paste the code correctly into the text box.)
If the question specifically asks for help to get the code to compile, fix a bug, or add a feature, then it's off-topic. Please don't answer it, as it limits our ability to regularize the question.
Ideally, we would close it, let the author fix the problem, then reopen it for a code review. Once an answer that reviews the broken code has been posted, though, we can't do that anymore. The question stays in limbo forever, which is not good for anyone.
Otherwise, if the code is broken, then it requires a judgement call.
Is it obviously broken, as in "you obviously didn't test this code at all"? If so, it should be considered off-topic. Close/flag it, leave a comment, and don't answer.
If it's plausible that the author was unaware of the brokenness, then it's a fine question. One of the purposes of Code Review is to point out problems that the programmer didn't know about.
The point of closing a question is that it cannot receive any new answers. That's because it has problems that effectively prevent a useful answer.
As such questions with broken code can be considered questions that shouldn't be answered.
The point you raise here is somewhat understandable, and seems well intentioned, but there is a problem with it:
Answering a question sends the signal that these questions are "okay" for this site. The point is. They aren't. Code is rather binarily broken (or not). This is to simplify understanding and enforcement of that rule. Allowing "minor broken" on the site would make this a game of drawing a line in the sand.
That makes for a bad user-experience for any user. Those involved in moderation and those "moderated". Lines drawn in the sand are not transparent. The goal of close-reasons is the opposite.
Let's not draw lines in the sand and keep our simple "binary" rules, even though that model may be oversimplifying things
