Class diagram is used for Object oriented program like C++, Java etc. How can I represent it for C code? C code doesn't have class but they have functions. Each function have variables and other calling function. I want to represent the relationship between functions with variable and calling function. How to do that and what can we call that diagram?
Thanks~~
Thank you Peter for your brief discussion and suggestion.
Let me ask for your opinion for following matter.
What i am trying to do is make a model from source code and merge it with some other model(generated from specification) for testing purpose. Two things happen here: spec to model1 and source code to model2. Then model = model1+model2.
What if source code is erroneous? Then along with model2, model will be erroneous too. Is it valid to use that model for testing?
Hi Muhammad, as Peter put it, it looks like you are looking for a "call graph" or a "caller graph", depending on the direction of function calls.
Since you mentioned that you are trying to reverse engineer such graphs from existing code, I will add my 2 cents and point you towards Doxygen. It is a documentation generation tool that can generate call and caller graphs from code.
If I understand it correctly, you wish to compare two such models (let's call them graphs). One is what was specified and the other is what was indeed implemented. I find that such detailed specification is too rigid, except maybe for a public API. Internal implementations should be specified at a higher level, otherwise refactoring is difficult if not downright impossible. Flexibility is good for overcoming unforeseen problems quicker and easier.
http://www.stack.nl/~dimitri/doxygen/index.html
Thank Peter for taking time to reply this post.
Model means an abstract representation of a system. Usually in Model Based Testing a model is generated from the spec and using coverage criteria test cases are generated. So this model have no relation with the development process.
But i feels that that model is not enough. I need to include more information in the model to make the testing more effective. That's why i am trying to create something like a model from source code. It is not to be some model it means i with extract some information from source code and use those information with the model(it is from model based testing model). So it can termed as somehow reverse engineering.
You told "I would have thought that reverse-engineering a model for the erroneous code would have helped you construct tests that show it is erroneous ". Here what is happening is, erroneous code->erroneous model(by rev engg.)->erroneous test case. The main aim is to find error in the source code. My concern is does a erroneous test case confirm that there is no error in the source code?
The example you gave not concern with computational information. Specification doesn't contain computational information like data-flow, control flow information etc.
Although you made some point i need to think about. It really seems philosophical.
Thanks again.
Thanks Vasco for your reply....
Thanks for mentioning dodygen. I have worked with doxygen for generate call, class diagram etc.
But i do not want to compare to model but combine two model. As you stated correctly that One is what was specified and the other is what was indeed implemented. It is not necessary to be same kind of model. It can be two different type of model. Main thing is merge those two model and make a better model. From that better model " more efficient" test cases are generated.
You may want to have a look at AutoFocus, which is a modeling language with an accompanying open source tool chain (AutoFocus 3) capable of generating verifiably correct c code from an architectural (abstract) model of, e.g., a (cyber-physical) system. Whether re-engineering of c code is supported, I cannot remember (I doubt it - but, the formalism used may be a good starting point for you). AutoFocus has been developed at TU Munich in Prof. Broy's research group. Here is a short introductory technical report: http://download.fortiss.org/public/projects/af3/research/2010/TUM-I1021.pdf
You can Make DFD Data Flow Diagram or Called Bubble chart Bubble represent the function and you can put arrow as the input parameter to those functions
Just run Doxygen over your C code with a default configuration plus change the EXTRACT_ALL option to YES. You can do this in the GUI that comes with the installer.
http://www.stack.nl/~dimitri/doxygen/
It groups C functions by file.
For more diagrams, turn the CALL_GRAPH and CALLER_GRAPH options on.
During my beginner days of programming (late 80s), there wasn't any IDE. To quickly find/navigate to a function in a plethora of source files, ctags was (and still is) a godsend.
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration