Approach to solve bugs 
Author Message
 Approach to solve bugs

A bug can be solved in 2 minutes, 2 hours, 2 days or 2 weeks.
What's the best approach to solve a bug?
Sometimes, how long it takes, depends of how you try to solve it.
Anyone has a "generic methodology" to share.  Step 1, 2, 3, etc...


Wed, 18 Jun 1902 08:00:00 GMT  
 Approach to solve bugs

Quote:

>A bug can be solved in 2 minutes, 2 hours, 2 days or 2 weeks.
>What's the best approach to solve a bug?
>Sometimes, how long it takes, depends of how you try to solve it.
>Anyone has a "generic methodology" to share.  Step 1, 2, 3, etc...

In my experience, first check the
last thing you "fixed", no matter
how trivial or irrelevant to your bug.

Ask someone for help, or "step
outside the box" in newSpeak.
A fresh brain often spots the problem
right away.
--
Donald



Wed, 18 Jun 1902 08:00:00 GMT  
 Approach to solve bugs

Quote:
>In my experience, first check the
>last thing you "fixed", no matter
>how trivial or irrelevant to your bug.

>Ask someone for help, or "step
>outside the box" in newSpeak.
>A fresh brain often spots the problem
>right away.

The voice of experience!  Very good advice.  To that I'd add that Smalltalk
bugs tend to be a little different from other languages, and Simon Lewis
describes this nicely in "The Art and Science of Smalltalk".  Perhaps the
most important piece of advice Simon Lewis give is to "read the walkback".
Of course, one must always consider that the walkback might be showing
something that went wrong only because of some other "real" problem
elsewhere.  Still, there are times when it points directly to the problem;
either way, it's best to read what it says before closing it.

Good luck,

Bill

--
Wilhelm K. Schwab, Ph.D.
University of Florida Dept. of Anesthesiology



Wed, 18 Jun 1902 08:00:00 GMT  
 Approach to solve bugs



Quote:
> A bug can be solved in 2 minutes, 2 hours, 2 days or 2 weeks.
> What's the best approach to solve a bug?
> Sometimes, how long it takes, depends of how you try to solve it.
> Anyone has a "generic methodology" to share.  Step 1, 2, 3, etc...

I've read that Kent Beck gives the advice to "Let Smalltalk tell you what
the problem is." In other words, use the de{*filter*} to follow the code. It's
sort of the opposite of "read the walkback" and is a very effective
technique.

Doug Swartz
Gate Software



Wed, 18 Jun 1902 08:00:00 GMT  
 Approach to solve bugs


Quote:
> A bug can be solved in 2 minutes, 2 hours, 2 days or 2 weeks.
> What's the best approach to solve a bug?
> Sometimes, how long it takes, depends of how you try to solve it.
> Anyone has a "generic methodology" to share.  Step 1, 2, 3, etc...

Here's one generic sequence, which doesn't help much in any
specific circumstance!

1.  Determine the problem.  That's to say, convert "it crashed"
into "Error message xxx and walkback in method ...." or something
similar.
2.  Reproduce it.  This can be the hard bit, but until you can do so,
you can't be sure you either know the problem or, later, have fixed
it.
3.  Determine the problem source - find the bit of code which causes
the problem, which often isn't where the symptoms appeared.  For
example, a "Does not understand" error can be caused much earlier in
a program's life by someone leaving a value in a Dictionary set to
nil.  (And Java proponents of "strong typing" get exactly the same
problem, 'cept it's spelt "null"!)
4.  Code a fix.
5.  Use the test case produced in step (2) to prove the problem is
solved.
6.  Keep the test case for regression testing.
7.  If you have time, run all your regression tests.

Paul Gover
IBM Warwick Development Group
Opinions my own, not IBM's



Wed, 18 Jun 1902 08:00:00 GMT  
 
 [ 5 post ] 

 Relevant Pages 

1. Linguistic approaches to AI Linguistic Approaches to Artificial Intelligence

2. Linguistic approaches to AI Linguistic Approaches to Artificial Intelligence

3. To all CW5 users: MultiCopy Print and FileDrop Combo bug solved

4. Clipper 5.3bi bug SOLVED! (thanks to Murad)

5. 5.3bi bug - Solved? (to Stephen Quinn)

6. Optimization bug solved: thanks, Andrew!

7. problem solved, but I don't know why (was: GNAT implementation bug)

8. Watcom F77 linking problem solved (*BUG* report)

9. MS C/Fortran bug solved

10. tk8.1a1 bug with checkbutton and radiobutton menus solved (patch)

11. Have I found a bug? (solved)

12. I have solved/will solve AI!

 

 
Powered by phpBB® Forum Software