Jim

Yes 80% was a surprise.  But I agree with you on the
actual % not being a big thing as compared to what is
being covered.  jWebUnit will be goverened by the
80:20 rule as like anybody else.  ie probably 20% of
our code is the critical path.

So is the 80% coverage really the "ideal" 80% is the
question.  I think the answer is yes.  The tool
actually lets you drill down up the source code line
level.  So one gets to see what did not get covered. 
I liked what I saw (some getters, setters etc) 

In any case, the tool is open source so I will
integrate the whole thing and publish this weekend. 
That way, everybody can view the results.

Regards
Vivek

--- James E Weaver <[EMAIL PROTECTED]> wrote:
> 
> 
> 
> 
> Well,
> 
> That's a good percentage - a bit better than I would
> have expected.
> Application of a test coverage tool is nice -
> presumably it can drill down
> on the packages a bit more and show us where the
> misses mostly are.  The
> Pseudo Server borrowed from jWebUnit is a pain to
> maintain a copy of, but
> it did allow for pretty easy to read and thorough
> testing.  A simpler
> alternative to the pseudo server was another item on
> my todo list ages ago.
> 
> Good going vivek.
> 
> We just had a thread on a TW development list about
> "what percentage of
> coverage is good coverage".  Most leaned towards
> 90-something, but one of
> my favorite posts was this one by Mike Two (I don't
> think he'll mind me
> quoting him here):
> 
> "Attaching any particular number is a waste of time.
> Use your head, be
> pragmatic and measure coverage often. When you do
> measure it don't worry
> about the number, worry about the trends. Do some
> analysis to figure out
> what is or isn't getting covered and make some
> decisions about it.
> 
> Be afraid of rules like, "we must have 95%
> coverage". Measurement of a
> system changes the system. Since catch blocks are
> frequently mentioned as a
> problem and not so easy to test a lazy developer may
> start writing "catch
> (Exception)" so that they only have one untested
> catch block. Or moving the
> catch block higher up the method stack or writing
> long methods with one big
> try catch to keep the percentages of covered code
> up.
> 
> I've seen projects with 30% coverage and almost no
> bugs, I've seen projects
> with 90% coverage and a ton of bugs. Your tests can
> cover everything and
> still suck. Run coverage once in a while, ignore the
> percentage number and
> look for trends in namespaces and classes that
> aren't covered."
> 
> Jim
> 
> 
> 
>
-------------------------------------------------------
> This SF.Net email is sponsored by: GNOME Foundation
> Hackers Unite!  GUADEC: The world's #1 Open Source
> Desktop Event.
> GNOME Users and Developers European Conference,
> 28-30th June in Norway
> http://2004/guadec.org
> _______________________________________________
> Jwebunit-development mailing list
> [EMAIL PROTECTED]
>
https://lists.sourceforge.net/lists/listinfo/jwebunit-development



        
                
__________________________________
Do you Yahoo!?
Friends.  Fun.  Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/ 


-------------------------------------------------------
This SF.Net email is sponsored by: GNOME Foundation
Hackers Unite!  GUADEC: The world's #1 Open Source Desktop Event.
GNOME Users and Developers European Conference, 28-30th June in Norway
http://2004/guadec.org
_______________________________________________
Jwebunit-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jwebunit-development

Reply via email to