MAXIMUM SIZE OF APPLICATION

[cyrille de brébisson]

Hello,

I do not remember exactly the flow of operations... But basically, it goes like this.

IO layer receives the compressed packet.
IO layer advises the bottom layer that a packet is ready
bottom layer decompress the packet (But since the packet bellongs to the IO layer, it can not free it at this point). So we have a *2 memory at this point.
Then the packet needs to be "interpreted" and the data saved.
As each blob of data in that decompressed packet gets managed, it will be copied to it's location, which might be in RAM, and even if it does not reside in RAM, caches and the like will keep a copy, making it a *3 at this point.
Then the decompressed data blob can be freed.
Then the upper layer returns to the IO layer which will free the packet

And we are done... but they was a temporary *3 on memory use :-(

Could we do better than that?
Yes of course, but it would be much harder to implement and therefore take more time and effort and this time would be taken from other activities...
It's a question of priorities and allowing large apps is not our priority at the moment.

Cyrille