Path: utzoo!utgpu!news-server.csri.toronto.edu!helios.physics.utoronto.ca!physics.utoronto.ca!petersn
Newsgroups: comp.sys.apollo
From: petersn@physics.utoronto.ca (Mike Peterson)
Subject: Re: <None>
Message-ID: <1990Aug1.144400.26957@helios.physics.utoronto.ca>
Organization: University of Toronto Chemistry
References: <9008011411.AA10103@richter.mit.edu>
Date: 1 Aug 90 18:44:00 GMT

In article <9008011411.AA10103@richter.mit.edu> krowitz@RICHTER.MIT.EDU (David Krowitz) writes:
>They would want to use Sun RPC in order to run the gobs of software which
>uses Sun RPC calls (like, say, PC-NFS, something which Apollo *still* does
>not support to my knowledge). Why is it that Apollo, and now HP, can't seem
>to get it through their heads that customers want, *need*, and *MUST* be able
>to use software packages that don't necessarily come from the vendor who
>provided their hardware? Many times we have NO choice as to the software we
>must run -- a collegue sends us their program which they developed on their
>Sparcstation and we can either
>a) spend 6 weeks of our valuable time to re-create the program on the 
>   Apollos rather than writing our papers/PhD thesis.
>b) buy a Sun workstation to run the software.
>
>Our research grants don't pay us to port software. They also don't pay us
>to publish papers 6 weeks after the same results have been reported by
>someone else. Which option do you think we would take? (I'll give you
>a hint: Since 1988, roughly 6 new professors in our department have
>bought workstations. None of them bought Apollos. In the past 12 months,
>5 of the 8 professors who own Apollos have bought brand new color 
>Sparcstations -- 6 machines at roughly $18,000 each after discount -- and
>only 2 of 8 professors who own Apollos have bought Apollos -- 5 new DN2500's
>at roughly $3500 each after discount -- that's roughly a 6 to 1 ratio in
>dollars spent).

I agree with these comments completely -- I have spent (and continue to
spend) ridiculous amounts of time porting BSD programs that should run
on any BSD system without change (e.g. sendmail, named, ckermit,
batchd (still not working), C-News/rn/nntp, NCAR library).
We have also had no end of problems with TCP/IP and Ethernet, especially
on our DSP10020 - more on that later, but it still doesn't work for more
than 1 week average (we are running SR10.2.0.5.p)..
No one else at our University even thinks of HP/Apollo after our
problems - who needs the headaches? We are going to buy SGI or IBM
RS/6000.

Mike Peterson, U/Toronto Chemistry,
system@alchemy.chem.utoronto.ca		(416) 978-7094