why is it impossible for testing to find all bugs in a system?

Question

why is it impossible for testing to find all bugs in a system?

Answer 1

Yes it is impossible. There are so many permutations. Different operating systems, 32 or 64 bit, different computer brands, different hardware is some cases, even different languages for the operating systems. Trying to run all the tests on each permutation is not cost effective. And that is just finding a bug. It is another story when it comes to taking the expense at fixing them. You try to fix the high and medium impact ones.

For example, Windows 98 in Russian had a defect with the Windows calculator. When changing views, the view would collapse making it impossible to operate it. And that was Windows 98 or Windows 98 Second Edition in Russian. The developers did not believe me until they ghosted one of their machines to it and recreated the defect. It was a business reqirement that the users of our software could pull up the calculator. Ruled a minor defect with low impact and never fixed.

Answer 2

Even on a single computer, with 10,000 lines of code (that's a small program) and thousands of different possible inputs, it would take years to test every possibility - by which time the program is 3 versions newer and testing the old version really silly. We had a program that included a database. If the color entered in one table of the database was avocado, there was a bug. No other color. It took 6 months to find out that much (the program kept failing for no apparent reason). In over a million lines of code, you can see how long it would take just to find out what the bugs are, let alone fix them.

Answer 3

What is the evidence of this assertion? Since the entire basis of your theory is not just an assumption but a series of assumptions that build upon one another, you will need something tangible to back this up. Oxygen, at concentrations necessary for life to flourish, was present in the atmosphere during Creation week at the beginning. The geological evidence, including sulfur minerals and carbonate rocks, is explained by deposition during the early part of the global Flood. This same flood buried all of the animals and vegetation and formed rapidly the fossils we are uncovering today. Also, oxygen is a very reactive molecule, and aerobic creatures must have means of dealing with it. Hence the important role of anti-oxidants. It is therefore not surprising that too high an oxygen level can be toxic. So is not clear that higher oxygen concentration will be beneficial. Regarding larger insects and O2 levels: the next step would be to examine the tubes in insects found in amber, as a possible indicator of past oxygen levels. It certainly seems that this [test] would be much more definitive than trying to judge the pre-Flood atmosphere O2 concentration based on analysis of air bubbles trapped in amber. Amber is unlikely to form a seal impervious to gas molecules, and bubbles add to the pressure in any case. Whereas the tracheal tube comparisons could conceivably tell us about the oxygen content in the atmosphere in which the insect actually grew to maturity.

Answer 4

If you set out to test a system of a reasonable or more complexity, it would take too much time to design tests and execute them all to find all the bugs in the system.



I suppose that it should be possible to mathematically prove that you may never find all the bugs in a sufficiently complex system.

Answer 5

Well nothing is perfect. The cost involved in trying to achieve a system which is 100% bug free the first time is extremely high. If that cost is forwarded to the customer the system will never be built.

Answer 6

!/usr/bin/perl -w



print "hello, World\n";



=================



I have tested and certified the above program works on my Redhat Linix.



WARNING: it may not work on your installation if:



1) You don't have Perl installed.

2) Perl is installed in a different path than "/usr/bin"



Add another 100,000 lines or so, and things get slightly more complex - and unreliable. And... who has time to test 100,000 lines of code for EVERY possible issue?



Enter things like "Risk Based Testing." Portions of your application MUST work or the application is worth a hill of beans. Those portions will be tested most. This leaves those portions least important tested less. Bugs will be found - by user's - once the application is deployed.



EDIT: Seems Yahoo removes the begining sharp (#) on the first line of my program. Oops - the program, if you copy it to YOUR installation, will now have an error in it. Please add the sharp (#) on line #1. THANKS! :)

Answer 7

Bugs are made to be really smart and so even some of the top testers cannot detect them so they can continue to be in your system.