cvs commit: src/sys/geom/concat g_concat.c src/sys/geom/label g_label.c src/sys/geom/mirror g_mirror.c src/sys/geom/raid3 g_raid3.c src/sys/geom/stripe g_stripe.c

[M. Warner Losh]

In message: <20040826201531.GQ77326 at green.homeunix.org>
            Brian Fundakowski Feldman <green at freebsd.org> writes:
: On Thu, Aug 26, 2004 at 10:09:16PM +0200, Poul-Henning Kamp wrote:
: > In message <20040826200728.GP77326 at green.homeunix.org>, Brian Fundakowski Feldm
: > an writes:
: > 
: > >> They have been very successful in this respect, and more will be
: > >> added as I think them up.
: > >
: > >Why in the world would you crash the machine when you don't have to?
: > 
: > To prevent bogus code from being committed in the first place.
: > 
: > Presumably if people can not make their code work because it panics
: > they will not commit it.
: 
: You cannot presume testing will expose every possible state/condition.
: It hasn't in FreeBSD's past, but maybe you can start a revolution that
: makes it simple to create very complete test suites and put them into
: the tree along with the code being tested.

I beg to differ.  newbus has lots of asserts in it that could be
recovered from, but instead are panics.  These panics have been very
helpful in shaking out bugs (much to the annoyance of the person doing
the work until the bugs were figured out).

Warner