[Ns-developers] Multiple tests in regression python script.

[Mathieu Lacage]

On Wed, 2008-06-25 at 13:19 -0700, David Evensky wrote:
> 
> How do I run multiple tests within a regression python scripts?
> 
> I have currently something like:
>     import os
>     import shutil
>     import tracediff
> 
>     def run(verbose, generate, refDirName):
>     """Execute a test."""
> 
>     return tracediff.run_test(verbose, generate, refDirName,
>                               "tcp-2way --createTraces")
>   
> but I also want to run with multiple things like:
> 
> tcp-2way --createTraces --packetGap=31 --useDummyBytes=0 --recvSize=400
> 
> Given that the regression test will be testing multiple things, should
> they all create tracefiles? (and if they do, should they all be appended?)

I don't think they should be appended. It seems simpler to separate
them.

> 
> What sort of output should the regression test produce?

The python script should just output a PASS/FAIL statement. The
regression test itself should output whatever we want to ensure is the
same across multiple runs so, the more, the better, unless, of course,
you generate gigabytes of traces we have to store in our trace repo.

Mathieu