[interfacekit] Re: Ready, set, go!

Good to hear from you.

I'm inclined to agree with you re. using cppunit with multiple threads.  
Extending cppunit to deal with threads explicitely could be really good 
for the entire OBOS team, so I would definitely encourage you to go 
ahead with that.

e

>Its been a while since I have sent an email also.  Between holidays and 

>some personal things going on, I have not really had a chance to work 
>on anything over the last 1.5 months or so.  But I plan to get back at 
>it now.
>
>The things I am going to be working on are:
>
>- documenting and writing test cases for Autolock (I needed it to test 
>BMessageQueue so I implemented it quick)
>
>- re-implementing the BMessageQueue and BLocker testcases using 
>cppunit.  See below for some details
>
>- document and implement BMessageFilter class, assuming it is still 
>available.  Seems to be according to the webpage
>
>On the topic of cppunit, I looked at the docs and I think I may have a 
>problem using it with the testcases I have written.  The problem is 
>that to test BLocker or BMessageQueue well, I had to do tests where a 
>couple of threads are started concurrently.  These threads attempt to 
>find synchronization problems by using the interface concurrently from 
>multiple threads.
>
>I am unsure how well cppunit will work with multiple threads.  My 
>understanding is that a failure of a test turns into a C++ exception 
>being thrown.  So, in my situation the stack might look something like:
>
>main thread - RunTests(), blocked waiting for threads to finish
>thread 1 - doSomething(), performing the test
>thread 2 - doSomethingElse(), performing the test also
>
>If something goes wrong in thread 1 or thread 2, how will the exception 

>make its way to the main thread so it can be reported as a test 
>failure?  Correct me if I am wrong, but I think we need to implement a 
>class derived from TestCase called ThreadedTestCase perhaps.  This 
>class would provide a mechanism to start a set of threads, each of 
>which perfoms part of the test, waits for them to finish and reports 
>any failure.
>
>What do people think about this?  Is this a requirement we need to 
>build ontop of cppunit or is it there already and I am missing it?  I 
>will volunteer to do so if people agree it needs to be there.
>
>Again, sorry for the lack of any real progress from me in a while.  I 
>hope to change that very soon now!
>
>--
>Jeremy Rand
>jrand@xxxxxxxx

Data is not information, and information is not knowledge: knowledge is 
not understanding, and understanding is not wisdom.
        - Philip Adams

Other related posts: