Re: When is a bug a bug?

["Steve Landess" <sjl_abc@xxxxxxxxxxx>]

On the topic of bugs...

Ed McVaney of J.D. Edwards for a time had a newsletter that he called the 
"Rolling Letter".

I recall one particularly amusing issue from the late 80's - early 90's, 
where he was discussing bugs in the JDE package.  In an attempt to make a 
point that JDE had a low number of bugs, he described it using a formula 
that shows the number of bugs per line of code in the package.

His formula:
Number of bugs per line of code = (Number of known bugs / Number of lines of 
code in the package)

His specific example of this was something like:
"As of this edition of the Rolling Letter, we have approximately 440 known 
bugs, and there are approximately 10,000,000 lines of code in the entire JDE 
package."

His resulting statement in the Rolling Letter...
"So, the current number of bugs per line of code = (440 / 10,000,000) * 100 
= .000044 or there is currently .0044 % bugs per line of code in the JDE 
package."

This was highly amusing to me, since this is simply an illustration of the 
old Mark Twain saying:
"There are Lies, Damned Lies, and Statistics..."

Why?
Because JDE, like many other vendors, will not always take credit for the 
TRUE number of bugs found in their package.

Over a period of approximately 15 years, I almost NEVER got JDE technical 
support to admit that they had a bug in any of their code - they would 
typically respond in one of four ways:

1)  Have you modified the code?  If so,  then YOU obviously introduced the 
bug when you modified the program.  (I learned to answer this question NO 
after the first call - I would first perform a test to determine if the bug 
I was reporting existed in their original pristine code BEFORE I called to 
report it - and in almost every case it DID!)

2) We can't duplicate the error that you described on OUR system (therefore 
it MUST NOT exist!)

3) The program is working as designed (Answer given MOST often!)

4) We have identified this problem (a very rare occasion indeed), and it 
will be fixed in the next release.  If you can't wait for the next release 
or you are rev-locked, you can apply a "paper" fix to your code, but we 
won't guarantee it!.


SO, to make a short story long, the true number of bugs in the JDE code was 
probably more like 4400 than 440, since MOST of us had given up on 
attempting to get JDE to fix the problems - we just fixed it ourselves, if 
we had the source code for the program.

Regards and LOL,
Steve Landess
Austin, Texas
(512) 423-0935