Position & proposal

Mike McDonald mikemac@titian.engr.sgi.com
Wed, 04 Jun 1997 14:24:58 -0700

>From: Paul Prescod <papresco@calum.csclub.uwaterloo.ca>
>Subject: Re: Position & proposal
>To: mikemac (Mike McDonald)
>Date: Wed, 4 Jun 1997 16:35:04 -0400 (EDT)
>>   Adding multiuser capability causes all sorts of headaches concerning
>> who can do what? when? who does it affect? when does it affect them?
>> For instance, you're off modifying some system service. Do the other
>> users get your new modified services or do they use the old service?
>I think that if it is a shared service, they get the new service. If it is a
>service that is non-shared, they keep their version and you keep your 
>version. You don't need multiple users to run into these deliemmas -- you
>get them as soon as you move into multiple concurrent processes. What happens
>when you change your mail spool process: do all of the daemons concurrently
>working with it use the new one or the old one? These daemons may not be
>officially "users", but they introduce all of the concurrency problems of 
>concurrent users.
> Paul Prescod

  The difference is that when I mess with things, I'm willing accept
the possibility that I'll break things along the way. I don't expect
the system to "magicly" break when I'm just using it. Say you're trying
out some new ideas for the POS but I just using it. I use your new
version to save my irreplacable data. (It's always worked in the
past.) But you have a subtle bug in your new version that you are
trying to debug. I'm hosed because you can't try out your new version
without forcing me to use it for production work. Nasty.

  Mike McDonald