Re: Milestones document

[Havoc Pennington <hp redhat com>]

On Tue, 7 Dec 1999, Michael Moore wrote:
> Hi, this is all very new to me so your tolerance is appreciated.
> I read the web pages indicated below, but there is something that
> is still not clear to me. Specifically, what is SHARE. 
> I mean, is that reenterant code or is it a common (shared) pool of
> variables?
>

It's stuff in shared libraries, such as GTK or the GNOME libraries.
This data is shared between any apps using the same libraries.
 
> And why is RSS - SHARE appropriate? For example, if you are running a
> single program-A whos RSS = 1000 and SHARE = 750, then according to the
> recomended evaluation of the amount of memory being used would be 250.
> But suppose that this is the only program running. In other words,
> there is the POTENTIAL to share 750, but in actuality there is no other
> job/program that actually is sharing the 750. Wouldn't it then be more
> appropriate to say that program-A uses 1000.
> 

Yes, but in a GNOME context typically you have quite a few programs using
the same libraries.

Another interesting way to look at memory is the "memprof" tool; it will
tell you how much _allocated_ memory a process is using (which is unshared
memory not on the stack).

Havoc