8.10b - first impression and questions

[Knapp]

On Wed, Oct 8, 2008 at 3:57 PM, Derek Broughton <news at pointerstop.ca> wrote:
> Knapp wrote:
>
>> On Wed, Oct 8, 2008 at 11:55 AM, Eberhard Roloff <tuxebi at gmx.de> wrote:
>
>>> I have yet to see perfect software. Even Linux does not offer it and
>>> imho nobody will ever be able to write a perfect solution.

I took this to mean bug free and not that it would work when subjected
to things it was not designed for. Perhaps a failure on my part to
understand you.

>> Really there is a lot of perfect software, just not much for the home
>> user because it costs to much to produce.
>
> Really?  Anything more complex than "Hello World" is likely to have errors.
> Even mission-critical, lives-at-risk, software was found to have errors
> during the Y2K fiasco (not that it was likely that many of those errors
> would have broken anything - but the software was not perfect).

There is a difference between perfect and bug free, I guess. Y2K bug
existed because the first programmers never imagined than their
software would still be being used in the year 2000. That is not a bug
but a piece of software that is being used passed it's expiration
date. If you want to look at this way then the qwerty keyboard is a
very bad bug. :-)

>> What is not perfect on the
>> home systems is often the hardware and other peoples software that it
>> must interface with anyway. When you release a piece of software it
>> must not work with just one type of hardware but 1000s! It is really
>> impossible to test 1000s of possible hardware types that you might
>> have your software run on and it is not cost effective ether. Zero
>> fault software is the sort of stuff that NASA uses.
>
> Like the lander software for Mars missions :-)

Except that was not a programming bug but a data entry bug because the
USA is so stupid that they can't teach their scientists metric. (I am
from the USA btw and resent that I was not properly taught the
international standard ether in any useful way.)

Yes, even in zero tolerance environments mistakes do happen. It is all
a matter of how hard you want to look for the bugs. If you pay 10
people to look over every line and a 10 more to test every line, you
should end up with a perfect program that cost you millions to
produce.

One way that MS beat the rest of the world was to stop looking for
bugs so hard and thus make their programmers much more productive and
their software much more feature rich. For some reason people excepted
the bugs for a long time without much complaint as long as they go the
cool features. MS got their software made cheaper and faster than the
other companies and won. I will not get into all the idea stealing and
buying that went on.

As for perfect software, I think the Linux kernel does a very good job
at this. When was the last time you had a kernel crash on a Linux
certified computer? Maybe not, "Perfect" but darn good.

-- 
Douglas E Knapp

http://sf-journey-creations.wikispot.org/Front_Page