Thou shalt have no other gods before the ANSI C standard 1650

It's a silly argument, because nobody in their right mind would use sizet for counting something like that.

That is not the data type's fault. The programmer made an buttumption that whatever sizet was on his platform would be big enough to hold the largest amount of data that might come through the network pipe. sizet is not supposed to be "the largest possible counter".

The programmer made a stupid mistake. If the point was to count any amount of data that came in accurately, an extended precision integer extension, or manually handling overflow and storing across a suitable number of longs would have made much more sense.

I wrote a stress tool a few years ago to test data integrity over various fibre channel and gigabit ethernet adapters and switches. It had performance counters, and it didn't overflow because I decided to buttume that 32-bits of counting was all that would be needed.

I doubt anyone on an 8-bit microcontroller would be dumb enough to think sizet would be ok for this, but 32-bit programmers are typically dumb enough to buttume just about anything. The larger the address space, the more laziness sets in. Now we already see that 64-bit is seen as a brick wall that will never be hit.

-- Randy Howard (2reply remove FOOBAR) "Making it hard to do stupid things often makes it hard to do smart ones too." -- Andrew Koenig