Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!emory!hubcap!ncrcae!nncrcae!ofc!rogers
From: rogers@ofc.Columbia.NCR.COM (HL Rogers)
Newsgroups: comp.sys.ncr
Subject: Re: NCR Multi-processor unix
Message-ID: <1991Apr19.111434.2439@nncrcae.Columbia.NCR.COM>
Date: 19 Apr 91 15:14:34 GMT
References: <651@wrangler.WLK.COM>
Reply-To: hl.rogers@ncrcae.ColumbiaSC.NCR.COM (HL Rogers)
Organization: NCR Corp, E&M-Columbia, Columbia, SC
Lines: 21
Nntp-Posting-Host: ofc

In article <651@wrangler.WLK.COM> nolan@helios.unl.edu (Michael Nolan) writes:
>In a sidebar, the article states that applications developed in SVR4 'will
>run unchanged' on SVR4 MP, providing that the developers write portable
>code, meaning using standard system calls.  
>
>Question:  Does that mean that there will NOT be a common ABI between
>SVR4 and SVR4 MP, so that programs will have to be recompiled?
>
The section you reference in the article was using the context of a
customer who has been sold on the multiprocessing systems.  Since
these are not in production yet, NCR wants to insure customers they
can proceed with designing applications for the MP machines on their 
current SVR4 uniprocessor systems.  The cautionary statement perhaps
was confusing; it is a simple reminder to use standard calls.  
Neither the APIs nor ABI will change.  NCR's MP version is built upon
the uniprocessor version which is shipping today.
-- 
HL Rogers    (hl.rogers@ncrcae.ColumbiaSC.NCR.COM)
Me?  Speak for my company??  Why would I want to do that?
"Call 202/653-1800 for a good time!" - John Matrow, 1989
"The complexity is straightforward." - Doug Bartlett, 1991