On Wed 01 Apr 2009 at 12:10AM, Shawn Walker wrote:
> Danek Duvall wrote:
> >In the course of trying to fix my fix to bug 588, I kept running into
> >problems with the testsuite -- it would churn for about thirty minutes,
> >find a bunch of failures, and then crash, never spitting out any of the
> >output or tracebacks I'd need to find out what went wrong. It wasn't even
> >giving me test names, though I guess I would have had that at least if I'd
> >run with -v.
> >
> >So I whipped up an implementation of -vv which makes it dump out the error
> >log after each test (unless it passed), rather than at the very end.
> >
> >Is this something folks might find interesting?
> >
> >I also have a patch which adds a -l option to run.py, that makes all that
> >output go to a file, though I suppose it's just as easy to redirect it with
> >the shell. Again, any interest?
>
> I always end up running "pfexec make test-verbose" or "pfexec python
> run.py -o blah -v".
>
> Yes, yes, and yes. Almost every time I need to debug something that's
> failing in the tests, I end up altering
> cli/testutils.py:CliTestCase:debug so that it prints out _debug() output
> as the test progresses, which is very close to what you're doing with -vv.
>
> The -l is not so interesting to me, but the -vv is definitely.
You could just make -v do this, always. No need to add -vv.
-dp
--
Daniel Price, Solaris Kernel Engineering http://blogs.sun.com/dp
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
