[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [LTP] Testing Framework - An open source project in the works

I also like the distributed part of a distributed framework.  This 
allows for automated benchmark and stress testing of network applications.

Caldera is very close to open sourcing such a framework.  I have started 
the coding, but I am still in the process of finishing the complete 
design.  It will be able to do everything described, but will also have 
a modular design.  This will allow for a api to be writen, but also a 
pan module for quick move over for LTP test cases.  It will be written 
in Java, for portability, and the api that will make writing it easier. 
  Java makes it so that I won't have to make adjustments for different 
distributions and it will hopefully run on other platforms.  I am hoping 
that this will be helpfull, and usefull.

Jason Corbett

Urban Widmark wrote:

> On Thu, 19 Apr 2001, Michal ROSZKOWSKI wrote:
>> We have looked at a number of frameworks (including commercial ones) and none
>> of these meet our needs. I would like to know how much interest there is in
>> the community for such a framework.
>> I have written a short list of requirements at:
>> http://home.cern.ch/mkroszko/requirements
> I like the distributed part. I have a need to run tests that require
> changes on a server as well, both for setup and during the test run. With
> and without synchronization between client(s) and server.
> Being able to run the servers on different platforms is valuable to me.
> (for linux smbfs testing it is useful to run vs all sorts of win32
>  servers, unix/samba servers, OS/2 (?) and whatever else that can serve
>  smb ... win32 servers are most important for me)
> A test might be:
> 	linux client			win32 server
> 	+ write 10 bytes to a file
> 	  on the server
>         + signals the test-server
> 					+ write 10 more bytes to the
> 					  same file
> 					+ signals the client
> 	+ verify that the changes
> 	  become visible
> To create different cases (order of writes) some kind of signalling is
> needed.
> I suspect other cases are similar. To test network drivers you want to be
> able to run performance tests (a la netperf), or verify functionality vs
> different other OS:es/network stacks.
> pan has a very easy to use interface, write a program and the exitcode
> says good/bad. Specifying a C and shell API sounds more complicated to use
> for someone that haven't used this framework before. Perhaps that is
> unavoidable.
> /Urban

Jason Corbett
     _____     ___
    /  __/____/  / (_)__   __ ____  __
  /  /_/ __  / /__/ / _ \/ // /\ \/ /
  /_____//_/ /____/_/_//_/\_,_/ /_/\_\
==== /_____/ ===================================
             Caldera OpenLinux