>Programs should never request functionality without designing to make sure
>that there is enough memory for success. Or at least that is the followed
but didn't we all agree that pre-flighting is bad and unreliable to begin
with?
>Note that this is very different from running out of storage memory. This
>is beyond a developer's control so Palm OS does provide feedback and apps
>are expected to handle those cases. This should be tested too. One thing
ok, maybe this is just a gap in my knowledge. i've been assuming that if i
fill up my 2MB card with 1.999MB, then i have only 1k left over for the
dynamic heap. if you are saying that the 32kb (or 64kb, whatever) dynamic
heap is reserved and cannot be crowded out by storage memory, then ok sure
i'll buy that the OS doesn't need to be as careful about out-of-memory
conditions in FrmInitForm, etc.
but if the storage memory can crowd out the dynamic heap, then the OS has no
choice but to either handle out-of-memory gracefully, or just crash whenever
you try to do anything (such as launch the App->Delete... command from the
Launcher....). i would have to wonder how you could even delete things from
the device in order to free up memory.
so, there are two possibilities for Andrew's gremlins crash:
1. the dynamic heap gets crowded out by the storage heap and this is an OS
problem.
2. the dynamic heap is reserved and cannot be crowded out, and therefore
Andrew's app must be leaking dynamic memory and filling up the dynamic heap.
