> So what you're saying is that there's nothing analogous to the Win32
> translate exceptions mechanisms? This lets the OS exceptions (the beloved
> GPF et al) be translated into C++ exceptions. Otherwise Win32 exceptions
> (which are a quite different animal than their C++ counterparts) tend to
> fall through nasty cracks with nasty results.
Win32 doesn't force you to do C++ exceptions that way -- CodeWarrior for
Win32's zero-overhead exceptions don't catch OS errors, but also have
significantly less overhead in program code and data tables.
> If there is a way to catch various errors before they pop up dialogs, then
> it should be reasonably trivial to wrap these in a mechanism to translate
> into C++ exceptions. Now, the question is: Is there a way to intercept
> these errors before the system kicks into default behaviors?
You should be able to combine the system's ErrTry/ErrCatch/ErrEndCatch with
C++ exception handling. You'll just have to be aware that the C++ exception
handlers won't catch the system errors, and vice versa. However, this only
works if the system routine that originally sees the error uses the ErrThrow
system call. A quick scan through the PalmOS source code indicates that
ErrThrow is almost never used (but I can't really reveal more than that, due
to the Palm NDA).
Conclusion: use either mechanism, and be consistent. The system error
trapping calls won't trap real crash-worthy errors. If you're doing C code,
ErrTry et al looks pretty promising. If you're doing C++, use the
compiler's exception handling so the language will handle
constructors/destructors correctly.
--
For information on using the Palm Developer Forums, or to unsubscribe, please see
http://www.palmos.com/dev/tech/support/forums/
