I googled it and found http://www.bullseye.com/coverage.html, which explained what code coverage analysis was about. How do people do code coverage analysis in real world? Do they use commercial tools, open source tools, or in-house tools? What are the difficulty of using these different tools? It seems that some companies don't do code coverage analysis at all...
I would venture to say that most companies do not do code coverage. It is expensive and time consuming and requires development to build a special version for testing with all needed hooks in it. If I did it, and I have pushed for it several years now, I would go with McCabe's IQ. It's expensive to get set up, around $40K but I truly believe it could be worth it. There are open source tools out there but true Cyclomatic Redundancy testing is difficult to do cheaply, so I would question their effectiveness. Plus Cyclomatic Complexity was invented by McCabe.
Of course there is a bit of difference between coverage and redundancies, but it would seem that they are closely related.
Rational purify plus is a package available that allows u to do three things
1: code coverage
2: memory leaks
3: bottle necks in the code
by using this tool u can come to know how much of the code will be covered after the execution of the application.
in the sence if very less code is covered in a normal execution this is a indication to u that u need to do a review of the code .
if a lot is covered it means that ur program is slow
"if a lot is covered it means that ur program is slow
<font size="2" face="Verdana, Arial, Helvetica">You mean that a lot of unnecessary code is covered, some applications require a lot, (more than normal amount), of code to perform the function of that operation.
I have found that the most useful finding is when all tests are run, much of the code has not been covered, meaning the testing is inadequate or there is useless code in the application. But mostly the testing is inadequate!
There is another side to code coverage, which is the analysis of redundant code. This is meant for code that is actually executed but replicated due to a mid-80's delusion of "structured programming technique" that caused the same logic to be coded over and over again. From a maintenance viewpoint that creates a nightmare, and for testing you may have to test the same condition several times via a different context to make sure all instances do work as expected. You can find tools for all of the major programming languages to do an analysis but (in general) this is done mostly by coders.