Re: interesting byte.com article - SLS - paging

[jpcarr@xxxxxxxxxxxx]

>Folks, checkout
>http://www.byte.com/documents/s=7647/byt1035219761057/1021_woehr.html if
>you haven't already...
>Don in DC


Thank you Don

Great article/interview

<SNIP>

Byte.com: What would a serious, hard-core programmer want from OS/400?


FS: There are several aspects. If you look at the flat store, it's a
superset, in a very real sense, of Unix address spaces. So a few years ago,
we implemented something we originally called PASE, Private Address Space
Environment. We carved off a piece of single-level storage and made it
private to a process, which is a Unix address space. We carved off not just
one of those, but put in one million of those private address spaces within
single-level store. This allowed us to take Unix code and bring it right
over. At that time we picked IBM's AIX, but we've done the same thing now
with Linux. Our success is due to having these underpinnings that are a
superset of everything else.

OS/400 is becoming a controlling environment for multiple implementations
of Linux and AIX, and we also use it for Windows. We still run our Windows
implementation on an Intel processor, but the controlling part is really
OS/400.

Then there are persistent objects, which the rest of the world struggles
with, in terms of maintaining and garbage collecting these persistent
objects. Within single-level store we don't do anything like that. It's
designed to have persistence of objects, and there's no concept of garbage
collection because the storage is so large that when I destroy an object I
don't re-use the storage. It's this sort of programming paradigm that makes
OS/400 programming so pleasing at the theoretical level.
<SNIP>