Re: Re: One idea about Regression testing

Yes, Atul! We should do this kind of testing! I’ll take a look at FitNess. Even without it I think it wouldn’t be much of a problem to open some binary files, read data, parse it and write it back in a simple CppUnit test. Do you mean this as a means to test part of the networking subsystem or only the packet class (or both)? Maybe it is finally time to start to capture some of the requirements for Calitko and write acceptance tests that verify these were fulfilled. Putting data through the system would be such kind of test, right?

Regards,

Peter

atul wrote:One other idea is to use FitNesse ( www.fitnesse.org ) …

— cheers atul

atul wrote:

This is loosely based on the approach given in Practice Of Programming…

1. Let us capture a packet and dump it in as XML/text file

( this is a one time step ) — Generating packet fields with random numbers is also a very good idea…
2. Now read this file and construct the packet

3. Pass the packet thru the system and capture it back as a text/XML file

4. Compare this file with the one we have …

5. If the differ, we broke something

Do 2 to 5 for various files….

What do you all think?
— cheerio atul

Would you like to post a relpy?


This post is a reply to:
Re: One idea about Regression testing
One other idea is to use FitNesse ( www.fitnesse.org ) ... --- cheers atul atul wrote: This is loosely based on the approach given in Practice Of Programming... 1. Let us capture a packet (more...)

Follow-ups:
Re: Re: Re: One idea about Regression testing
Dear Peter It would be an end to end testing.... You are right about putting data thru the system and verifying that it still works... And auto verifying that new changes (more...)