On Aug 20, 3:16 pm, phil oakleaf <n...@oakleafsoftware.co.uk> wrote:
If I have a program that compiles but crashes on 32bit but fails to
crash on 64bit then I would start by thinking it is connected (somehow)
to 64bit.
With what right do you expect any consistency when invoking undefined
behavior, such accessing bad pointers? That's your primary mistake. If
you don't understand, or fail to accept the simple truth of undefined
behavior, as defined by C and C++ programming languages, potential
difference between 64 and 32-bit environment is the least of your
problems.
(I, too have 64-bit system and work on my 32-bit code in it. It's been
~2 years now. __Not once__ did I notice a difference from a 32-bit
system WRT memory access, and I, too, produce my fair share of bugs. I
believe you are acting under hugely false presumptions.)
On Aug 20, 3:16 pm, phil oakleaf <n...@oakleafsoftware.co.uk> wrote:
Developing on a 32bit platform 99.9% of bad-pointer problems trigger the
debugger, I dont regard this as relying on luck but using the tool as an
assistant.
Well, you indeed have some cheek! You are relying on undefined
behavior (trying to, rather). How can this be called "using a tool" is
beyond me. ( And I know that 67.43% of all stats are made up on the
spot, so I doubt your 99.9 figure :-). )
Step up to the plate and provide a repro of a difference - same code
crashing under 32 and running (seemingly fine) under 64. Here's an
offer: if you do that, I'll take time to try to explain what happened.
There is no mystery in this (or any engineering) field, only our
inability to understand it.
Goran.