Ray Morris

+1 for Steve's proposal.

Here's how I think of it.  We want to see how many people "got it right". That's a different question from how many people "got it right the first time".    As grade can based on either the highest attempt or the last attempt, I'd think the same would apply here -if you're grading the last attempt, you want to know how each question affected grades, so we'd want to know how many people got the question wrong on the attempt that "counts" for the grade.  

Any question that is answered incorrectly far too often, bringing down people's grades, needs to be revewed to possibly improve the question.

And I don't feel I am "shooting people down" by expressing that opinion.

There is a big difference between discussing user interface design and going after people personally.  To me, this sounds more like "shooting people down" than discussing the best user interface for Moodle:

I find very disturbing the fact that some Moodle users want to remove a feature used by some other Moodle users just because they don't use it and so they see it as pointless.

Since there's no mention of user interface elements anywhere, and repeated comments about the person who mentioned the concern, some might see that as not so much a discussion of user interface design as it is a personal attack.

That comment about "shooting people down" might also reference the following.  Do you think this is about the interface, or are most of these words about the people involved in the discussion?:

Why would you make other people's questions invalid ? Of course when re-editing the question they just have to enter a non breaking space, but why would they have to do that just because you think that such questions should not exist ?

Maybe calling people sloppy will be productive:

Well, that's too bad. The fact that some users are not careful enough is no reason to remove a feature which is useful to other (more careful) users.  

Is that how you talk to people face to face at work, attacking their motives because they aren't familiar with your use case?

One major aspect of usability is to make it hard to screw up.  If nobody ever had an empty question on purpose, if it was always an accident, it would be good to adjust the UI to make that error difficult or impossible.  That's neither "removing a feature" nor is it "just because you don't use it".   Lack of that kind of input validation isn't a feature or anything else, it's the lack of something.  The proposal was to ADD a check, not to remove anything.  The proposal wasn't to add validation "because I don't enter empty questions", but to add validation if empty questions made no sense, if they were not useful to anyone.  A reasonable response is to say "I find empty questions to be useful", not to 

 Have you ever noticed on some bathroom stalls they have a little shelf or hook that folds down so you can set your package down, but you can't open the door with the hook down?  It may seem a strange design, but it's on purpose - you can't accidentally leave your item in the stall.   Similarly, on some ATM machines you have to retrieve your card before it'll give up the cash.  That way, you can't accidentally leave your card in the slot. That's called a "forcing function" and it's good design - when you can be sure that what you're preventing is something the user didn't actually intend to do.

That red highlight seems to be a good design.  I wonder if the OP can come up with another feature along that line - making sure the user knows that the question is empty.  (What some UI designers call visibility.)

(BTW a dialog asking "are you sure?" generally isn't what I'm talking about, because people click "yes" almost reflexively.)

Currently I don't see a way to have a quiz activity complete when all attempts are used.  One can set completion based on a passing grade, but I don't see how a failing grade can be handled well because of multiple attempts. You may not mark the activity as complete on view or on any grade because that might be the first attempt of three, for example.

Is there something I'm missing?  If not, I'll submit a patch to add a completion condition of "all attempts used" or something.

Edit - thinking about it more, what we really want, I think, is to consider it complete if they either pass or fail the quiz.  Passing is supported - that's just getting a passing grade.  Failing would mean getting a failing grade on all allowed attempts.  Since the OR of conditions isn't quite supported, though it's partially coded, I'm not sure how to proceed. 

I'm a little suprised - I would have thought others would have wanted to know when a student pased or failed, but I've not found any discussion in my searches.  We can't be the only institution that wants to tracks when tests are passed and failed while allowing more than one attempt, can we?  Am I missing something?