[thelist] General Statement - There's Bugs in Everything

Ken Schaefer ken.schaefer at gmail.com
Sun Oct 24 19:29:28 CDT 2004

On Fri, 22 Oct 2004 08:14:55 -0500, Rob Smith <rob.smith at thermon.com> wrote:
> One of the "big wigs" surprised me yesterday with a comment that "There's
> bugs in everything." I took this in lieu of web development:
> Although I whole heartedly agree with him, I open a discussion as to ...
> why?
> Is it out of laziness of not checking all the p's and q's?
> Is it inevitable because we're human and not everything we do is perfect?
> Is it ignorance because most everything we do in the web arena is "new?"

For starters, humans are fallible - we often make mistakes. We do it
in our everyday living, so I somehow doubt it's impossible to flush it
out of our coding. Obviously, as we go through life, we learn how to
avoid making common mistakes (well, most of us do), and likewise we
stop making common mistakes in our programming. We stop making syntax
errors, and progress (sic) to logical errors, and then finally to
design errors.

Secondly, we are imprecise communicators. What the end user wants
might not be accurately conveyed to the analyst, and then not
accurately end up in the specifications for the project. Add this to
time, energy and knowledge limits (for example, what end user
rigorously reviews every specifications document to make sure they're
getting what they want?), and you have a recipe for mismatched

Lastly, what someone might call a "bug" someone else calls a "feature"
(that might make you laugh!) or an "unsupported situation" or "outside
the specifications". I'm writing a word processor. I use a long to
store the current page number. Assuming an unsigned long, that'll give
me about 4 billion pages. The original specs probably call for a "page
numbering feature", but nothing that indicates whether this should
handle "negative numbers", "imaginary numbers", or documents than have
more than 4 billion pages.

So what happens when someone decides to be smart and start numbering
their pages at 4294967295 and the next page is page 1? Is that a bug?
Or is it a situation that is outside the specifications of the
application? I mean, we could handle documents with larger page
counts, but at what point do we stop? Obviously at the point that's
mentioned in the specs, but at what point do we stop writing specs? At
what point have we handled every reasonably conceivable situation in
the specs?

Someone on this list once said that applictions should never ship with
bugs - my personal opinion is that that particular person simply isn't
aware of the bugs in their applications - any non-trivial application
these days has "bugs" in it. Whether they are really coding errors, or
the result of other factors (eg misunderstandings or incomplete
communication between end user and developer) is another issue.

This is worth reading as well:
More on Quality 


More information about the thelist mailing list