Mea culpa. Actually, the example code I typed in doesn't make much sense in any case. I should have taken the time to produce a tested example.

I wasn't trying to disprove anything, just commenting in keeping with the thread. However, perhaps I should state my objection more directly. The transition from test-before to test-after is (to my knowledge) unique with this combination of do BLOCK and while. In every other case, the modifier is evaluated before the statement that is modified.

I understand (or at least I think I do), that this was done to provide for the pascal-style test-after loop construct. My problem with the decision to ban the while modifier on do blocks is that it is (IMO) the wrong way to fix the misconception that I so nicely fell into with my examples.

I think that the misconception arises because in every other case, substituting a do block for a single statement has exactly the effect of grouping its contained statements that one expects. So, it's the special case that causes the problem.

So rather removing what would otherwise be a useful construct, I'd prefer that the special case was removed.

As the loop keyword has been designated, would it be too confusing to use

loop{ 
    ... 
} while condition;
[download]

Don't feel the need to debate this further. I missed (and haven't found despite having looked for the original decision point a couple of times), the arguments behind the decision, but I accept that you've probably given this much more thought than I, and you usually get these things right.

I also fit in that camp of people that would like to use multiple statement modifers, but accept you have good reasons for your distaste for those also.