>You kill the app if it lets exceptions out...
That'd be great in theory. In practice, not all compilers and languages
support exceptions, and I'm unsure as to whether the OS should
mandate/enforce one way of implementing exceptions.
>Not even bothering to return
>an error code is akin to saying you don't think anyone that develops for
>the platform bothers to check for them.
You might be surprised what we find when compatibility problems are
reported to us :-) The truth is uglier than you might imagine...
regardless, we do have error returns for most functions, and the ones which
used to give fatal alerts mistakenly were just that, mistakes, and they're
being corrected. (Occasionally they might well have been overzealous
coding; I can't guarantee what was in everyone's mind at every moment!)
>How about actually giving the developer the chance to fix the problem?
I totally agree that a good number of the fatalAlerts which existed in 3.0
and previous didn't need to be there. At the same time I think that some
number (small, perhaps) do make sense to keep.
>I'll bet that none of the OSes you vaguely refer to behave in the same way
>as PalmOS...
Hmm? You've never seen a blue screen of death, sad mac dialog, or core
dump? What I'm saying is that there's a point at which any OS just has to
call a stop. That line was set too high, and we're lowering it, but I
don't think it makes sense to lower it to zero.
-David Fedor
Palm Developer Support