On Thu, 4 Mar 1999, Richard Hartman wrote:
> If you don't know how to handle it properly, there certainly isn't much
> benefit in doing random fixups, but there is just as certainly no
> benefit on continuing on as if the error had not occurred.
>
> If you get an error you don't know how to handle, the best thing to do
> is STOP. Notify the user. Do not pass "GO", do not collect 200 more
> errors.
Actually, what I went on to describe was a technique that allows precisely
that to happen, but with a reduced burden on the programmer due to more
cooperation from the OS, and without as much ambiguity-of-purpose as C++
exceptions. Just MHO, of course, and certainly no excuse for someone to
skip the error checking in their PalmOS code.
Come to look at it, though, Palm doesn't seem to go overboard with the
error checking, either. This seems to come down to whether it is actually
worth spending the time and program code space needed for every possible
error check, especially as half the errors essentially can't occur unless
the OS has broken somehow -- Palm's code doesn't check for lawnmowers, in
other words.
Then again, that's not a good example to use: they wrote the OS, so they
might have a better idea were the lawnmowers are hiding...
--
Kenneth Albanowski ([EMAIL PROTECTED], CIS: 70705,126)
