home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.software-eng
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!ames!agate!rsoft!mindlink!a3923
- From: Gerry_Furseth@mindlink.bc.ca (Gerry Furseth)
- Subject: Re: Will we keep ignoring this productivity issue?
- Organization: MIND LINK! - British Columbia, Canada
- Date: Wed, 18 Nov 1992 05:06:46 GMT
- Message-ID: <17556@mindlink.bc.ca>
- Sender: news@deep.rsoft.bc.ca (Usenet)
- Lines: 37
-
- |I recall my first software class. Here is Fortran syntax; write a
- program.
- |Not much training. Not much experience. Compare that to the first _TWO_
- |years of hardware.
-
- My experience in the land of software has been entirely similar. I have
- found, though, that my EE courses come second to thermodynamics (but still
- ahead of computing courses). For those of you who skipped thermo in favour
- of more relevant courses, the basic precept is that every analysis begins
- with:
- 1: what is the best possible behaviour of this system?
- 2: what is the worst possible behaviour of this system?
- 3: how near to the best case can I realistically expect?
- 4: what information do I need to solve this problem?
-
- The last question seems obvious and unnecessary at first glance. But, after
- three semesters, my point of view had changed. As an example, an essential
- piece of information in evaluating the efficiency of any fuel powered engine
- is the ambient air temperature. Not always the first thing that leaps to
- mind.
-
- The thermodynamic analysis ends with: what assumptions are hidden in my
- solution?
- For a computing example this time, "I assume that no one will want a file
- larger than 1Gb." This might strike a chord with DOS developers: "I assume
- that a single-user operating system need not be re-entrant."
-
- The level of discussion on this topic has been consistently high; perhaps
- someone can restate what I mean more compactly and argue for or against.
-
- Aphorism: If you can't write a testing plan from your design, the resulting
- code will be unreliable.
- Collorary: If you write and execute a testing plan for all your designs, you
- are not working in software.
-
- All (my) opinions are no more reliable than the author.
- Gerry.
-