Name: dkR10031 Date: 06/25/2003



Current implementation of the Jcov tool expects that
result coverage data must be saved to the one result
file. However, instrumented application may be very
large and complex and saving of the result coverage
data to one file may be impossible since file
system, or operational system, or security restrictions.

Another reason to distribute result data is following
situation: any user don't know class which should
be analyzed for coverage before running or before the
end of execution of application and application is
very large (~10 000 classes). For current implementation
user should not use any filter and should gather all
coverage data to a very big file. After that he should
load this big file to the GUI and should work with all
coverage data although he already know which one class
should be analyzed.

It makes sense to distribute the result coverage data
by the following way: save coverage data to the files
which have names constructed by adding the extension '.jcov'
to the class names (for example, the coverage data file
Test.jcov is for Test class). In this case for the
described situation, user can find needed jcov file
using standard 'find' command line utility and load
to the GUI only needed coverage data.

Also it makes sense put coverage data for the class
to the directory there classfile with this class is
located.

======================================================================