[PJUG Javamail] debugging server-side code
bill.jackson at oracle.com
Sat Aug 11 19:57:01 UTC 2007
+1 to Matt.
Steven, I can't help being curious about the details of your bad
experience; I have never had such problems (though I have only been
using a live debugger for the last four years). I just start my server
or app with the JDWP RI agent enabled, and off I go. I even build it
into all my Ant scripts, using optional properties with reasonable
defaults (i.e. suspend=n, no address).
It seems to me that the debugger is just another tool in the developer's
toolbox, and an absolute refusal to it can be just as harmful as
over-reliance on it.
Matt Reynolds wrote:
>> I'm not kidding, either. I haven't used a debugger in years, except
>> once or twice as a final last-ditch tool after trying everything else.
>> Once, several years ago, I jumped through all the hoops required to
>> try to get my IDE to remote debug an app server using JPDA. That
>> horror pretty much ended my debugger dependence.
> I've found getting Eclipse and newer JVMs into "debug mode" to be a
> breeze. Setup the remote VM to wait until a debugger attaches, and
> away you go.
> I'm normally all about not using the debugger, but the place I've
> found it's needed most is with legacy code (which you almost always
> inherit at every job) that has poor unit testing. Worse if it's
> tightly coupled and resistant to unit testing.
> I've also found that IoC complicates the matter if it's not organized
> in a clear manner. I have an interface, but I don't necessarily know
> which class is implementing it without adding alot of debug statements.
> If you're writing something from scratch, have a well designed app,
> and have unit tests, don't use the debugger. Unfortunately, that's
> not been the norm at the places I've worked.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Javamail