Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!pyramid!voder!tolerant!kennedy
From: kennedy@tolerant.UUCP (Bill Kennedy)
Newsgroups: comp.unix.microport
Subject: Re: Memory models?
Message-ID: <1358@tolerant.UUCP>
Date: 1 Mar 88 18:04:52 GMT
References: <4022@umn-cs.cs.umn.edu> <473@spectrix.UUCP>
Reply-To: kennedy@tolerant.UUCP (Bill Kennedy)
Distribution: na
Organization: Tolerant Systems Inc., San Jose
Lines: 26
Keywords: 386 286
Summary: File formats too

In article <473@spectrix.UUCP> clewis@spectrix.UUCP (Chris R. Lewis) writes:
>In article <4022@umn-cs.cs.umn.edu> randy@umn-cs.UUCP (Randy Orrison) writes:
>>I have noticed people posting in here talking about programming and using
>>various models.  Is this for real?  
>
>Unfortunately yes.  But only on a 286.  The 386 doesn't need this dreck
>in UNIX (386/ix, uPort).  All pointers are 32 bits on 386 UNIX (unless
>you're running 286 UNIX on a 386 - then you get what you deserve).

It's actually even worse than that.  It seems that there are different file
formats as well.  I have V/AT DWB and went to V/386 and it doesn't work
because V/AT DWB is in "STL" format which V/386 thinks means try to run it
with Merge...  So even though the binaries are allegedly transparent, there
are at least some that won't work.

The pain here is that a lot of stuff is either not yet available for the 386
or the 386 version is frightfully more expensive than its 286 kin.  Further
I have not seen a flood of stuff to take advantage of the 386 and the way it
is implemented in V/386, nor anything to ease conversion from working Vr2
code to Vr3 (read as "anything that worries about signals").  So there's
still good reason to stick with the 286 brain damage.  If you grew up with it
it's no worse than any other impediment.

The above are my own opinions, I'm a contractor at Tolerant, they were nice
enough to let me use their stuff so don't be mean and confuse them with me.

Bill Kennedy ...{rutgers,cbosgd,ihnp4!petro}!ssbn!bill or bill@ssbn.WLK.COM