Its a tossup between a small increase in time every TList expresstion, vrs A Big increase in time for a few exprestions.
I personaly always favour the faster the commabds you are likly to use in a main loop, the better.
In this case, its piddley easy peasy to extend TList and add a MyCount (or if Count isnt Final, just override that), and so I dont want a change to the core object for everyone.
I personaly always favour the faster the commabds you are likly to use in a main loop, the better.
In this case, its piddley easy peasy to extend TList and add a MyCount (or if Count isnt Final, just override that), and so I dont want a change to the core object for everyone.
but since Paposo has demonstrated that using an internal counter does not reduce TList performance
I have seen no demonstration, I have seen one were I:+1 is shown to be arbitary, but this does not include any fetching time, for what would be a field of a type, and not simply a local variable that is very probaly held as a register in the example given