[PJUG Javamail] debugging server-side code

[Matt Reynolds]

On Aug 12, 2007, at 10:28 AM, Ward Cunningham wrote:

> I'm with Steven here.
>
> The trouble with single-stepping is that it takes forever and when  
> you finally figure out what's going on you have no record of how  
> you figured it out. That means you're darn little help to your team  
> unless you just want to do their job for them too. (Have you ever  
> felt that way? Maybe its your debugger.)
>
> Aside: Here is a report on a debugger hack that actually left a  
> record. We used it to write some documents but eventually found  
> easier ways to make better records. http://c2.com/doc/case87.html
>
> Best regards. -- Ward

Again, not arguing that debugging is "The One True Way", but more  
that it's an appropriate choice given the system under development.

At the last few jobs, we've gotten around the "record" by pairing and  
talking about the harder solutions.  Although informal, at my current  
workplace, we tend to pair if a debugger gets fired up, simply due to  
the complexity of walking through the system.

That aside, how do unit tests capture "how you figured it out" if  
it's a systemic problem?  I'm curious how this knowledge is captured  
in the cases where there aren't debuggers used.  I suspect that the  
problem, in my case, isn't the debugger, it's the system (or possibly  
me ;) ).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://javac.com/pipermail/javamail/attachments/20070812/fc5f23e9/attachment-0002.html>