On 21 Jan 2005 at 1:23, A-NO-NE Music wrote:

> Just to be clear,
> Only the programmer knows how Finale is paging.  For example, on my:
> 
> Dual-450/1.5GB  37.54MB RAM, 267.98MB vm
> TiBook800/1GB   30.75MB RAM, but only 159MB vm
> 
> FinMac2005a (initial launch, one doc open) on OSX10.3.7 (cold boot)
> Both boot volume has more than 10GB free space for vm.
> 
> So, why less RAM is not acquiring for more vm as expected here?  Only
> the programmer knows.

The programmers of OS X, you mean -- Finale knows nothing about 
virtual memory, which is entirely handled by the operating system's 
virtual memory subsystem.

> By the way, for this testing, I opened the same Finale doc on my
> network at the same time, expecting Finale would give me warning on
> the second machine.  Nope.  It didn't.  I think this is bad.

I think the people who are blaming memory paging for the corruption 
have a fundamental misunderstanding of the structure of a modern 
operating system. To an application like Finale, there is *no* 
discernable difference between real RAM and virtual memory. The OS 
provides a memory pool and is entirely responsible for managing it.

Paging happens when the OS determines that an application needs more 
real memory (i.e., is swapped back into RAM) or is inactive and 
another app needs the real RAM more (pages it out to the swap file on 
disk).

Thrashing happens when conditions are such that the needs of the 
applications for memory are changing faster than the OS can complete 
paging operations, so that by the time a first paging operation has 
completed, conditions have changed and then require additional 
paging. Thrashing should happen only when there is a relatively small 
amount of RAM available and the applications running have aggressive 
memory requirements, including lots of memory pages marked as 
unswappable.

But thrashing should *still* not ever lead to the kind of corruption 
seen in the file overwrite bug, as this would mean that the virtual 
memory subsystem is mucking around with the contents of the pages it 
is swapping, which is something a VM would *never* do!

-- 
David W. Fenton                        http://www.bway.net/~dfenton
David Fenton Associates                http://www.bway.net/~dfassoc

_______________________________________________
Finale mailing list
[email protected]
http://lists.shsu.edu/mailman/listinfo/finale

Reply via email to