death of the locality

Adversary

I've never liked the principle of locality of reference. In the computing world where everything is so ordered and pricise (mathematically neway) they come up with a theroy that memory access tends to cluster.

Not really.

Sure if you use a lot of loops and the same functions over and over again you'll access the same memory over and over again but with the increasing use of object-orientated methods I find I often have code that looks like this :

myVariable = classOne->classFifeteen->classFour->classNine.array[x] + classOne->classFifeteen->classFour->classNine.array[y]

It might happen that all these classes and varibales are in the same block of memory but it's not very likely.

IMO as object-orientated programming becomes more widely used and systems more complex the principle of locality of reference will become less and less useful. Though what are the odds that it will be replaced (if at all) with something useful?

Hobbes

My cats name is mittens.

IMO as object-orientated programming becomes more widely used and systems more complex

When did you start writing that post? 1980?

Trojan

rotflmao

but seriously Hobbes, there's a lot of people still not on the oo bandwagon, take mainframe for an example.

Al.

Blitzkrieger

database prople say they'll never swith to it cos there's still not a standard.

Add to that I know 3 peeps in companies not using oo and I don't know that many peeps in software companies

Trojan

Well, of a very limited experience of 2 software houses, *both* used strictly non-OO C...

I know they weren't run of the mill avg co.s (one was a RAID production site, the other is mainframe), but still this shows that whilst OO is widespread it is not the be-all and end-all!

My $0.194324234923423 (on a pentium)...

Al.

[This message has been edited by Trojan (edited 13-06-2000).]