Hi.

Ok, I think, excluding the problematic methods is really reasonable in
this case. I did compile all my stuff with debug mode, but the JiBX
binder adds additional no-debug-mode methods. Excluding them should be
feasable, as these methods won't have any unit-test coverage and will
just mess up the results (I cannot test them because I don't know what
they are suppposed to do (and they are even not existant in the sources)
and it doesn't make any sense to test them because they are
automatically generated by JibX).
An other solution could be (a guy from the JiBX mailinglist recommended
that) to compile all files to a temporary folder, add the JibX stuff
methods, decompile them and recompile them with debug-on (this could
easily be done using the optional jad-ant-task would not require and
exclusions. It's just adding a few lines to my local build file.  But
still these methods won't have coverage and will just lower the overall
coverage).

Just tell me which solution you like better and  I'll do it.

Cheers,
Christoph


Philip Johnson wrote:

--On Friday, March 18, 2005 12:48 PM -0700 Tim Shadel
<[EMAIL PROTECTED]> wrote:

java.lang.UnsupportedOperationException: No line numbers detected in
org.hackystat.app.cgqm.implementation.TestMetric1.<init>. Either
remove the
'oneLineFile' tag or turn debug on when compiling.


Another option is to exclude that particular constructor from
coverage, using the excludeIndividualMethods option.

--Tim


I'd prefer that Christoph not do that.  We should exclude individual
methods when there
is a good semantic reason in terms of the analysis, not simply because
it's easier than
turning on debugging.

Cheers,
Philip



Reply via email to