From: Ankita G. <an...@in...> - 2008-05-22 13:07:35
|
Hi Gilles, On Wed, May 21, 2008 at 03:16:45PM +0200, Gilles Carry wrote: > Hello Ankita, > > Ankita Garg wrote: >> I like the idea. It helps making testing infrastructure to be generic >> and easy to extend. >> I might be missing something, or it could be that it is a TO-DO item, >> but the patch looks incomplete...the testcases dont yet support the '-c' >> option. Also, the parsers do not take the profiles into account. As I >> said before, this could just be your first step towards completing the >> entire infrastructure and you might have the rest in your To-Do.. > > Actually, the '-c' option is not mandatory for this patch to work since > the profiles can take any command line. > Arguments are not interpreted by the scripts which run the command lines > as found in the profile. > So from the code it looks like that the commandline arguments specified to the test in run_auto.sh are overidden by those in the profile. This seems to be the right behavior, as this would allow one to have, lets say two profiles, one each for jvmsim enabled and disabled. But this being the case, we should clean up the arguments from run_auto.sh. For example, run_auto.sh in sched_latency should now read: $SCRIPTS_DIR/run_c_files.sh $profile sched_latency as against, $SCRIPTS_DIR/run_c_files.sh $profile sched_latency -d 1 -t 10 Thoughts? > In a previous patch, I added that '-c' argument only to tests that feature > pass/fail criteria. Some tests don't feature it. This is why you don't > find this option everywhere. > Hey, I do not see the '-c' option being supported by any testcase...I am on ltp-full-20080430 though. The patches might have gone in after that ? > Does this answer your question? > > Which parser are you refering to? > I was refering to the log parsers available under the rt-test suite itself. But, I just noticed that they are present for the testcases that do not have a numeric pass/fail criteria (for eg, pi-tests). So, pl ignore. >> Otherwise, the patch looks fine to me. Applied it and seems to be >> working fine. > > Thanks for your feedback. > > Gilles. -- Regards, Ankita Garg (an...@in...) Linux Technology Center IBM India Systems & Technology Labs, Bangalore, India |