>>>>I think that these these two combos should be blank by default, but not be allowed to be blank when you press ssnd or preview, so that the users MUST make a choice. Did you read that, Michel?
>>>
>>>Not so sure. How much of version info do you need if there's something about commands which haven't changed in the last 18 years (like skip, count, or the whole arithmetic apparatus)? The version, OS and whatever are completely irrelevant, and furthermore, someone still on VFP5 or 6 may skip those messages as too tall for them, while they may actually benefit from reading them, or even contribute.
>>>
>>>I know, this doesn't happen too often, simply because there's nothing new in those areas, but still. And no, I have no clue how would Michel's app measure relevance.
>>
>>Anyway, I see no disadvantage. I have stopped counting how many times I have had to select which state I live in, even though Norway is not divided into states. The same goes for other irrelevant information, but in this case the version number or OS can in deed be very relevant.
>
>The programmer's validation conundrum. What is worse: irrelevant info, wrong info, or missing info? Since we don't necessarily go wrong here (except when we keep our default VFP version while asking something about colleague's problem, and he's using a different version), we are choosing between irrelevant and missing. As it is, we allow for missing; if we'd enforce validation-on-blank, then in some cases it would be irrelevant or superfluous. Maybe a combination of a mandatory entry and a "doesn't matter" as a selectable option?

Yes, that would be much better than the present solution.