On Fri, Jun 20, 2003 at 12:28:29PM +0200, Daniel Vainsencher wrote:
> 
> Marcus Denker <marcus at ira.uka.de> wrote:
> > On Tue, Jun 17, 2003 at 06:51:46PM +0200, Daniel Vainsencher wrote:
> > > And Marcus - 
> > > * Even in the failures we see here, we see a huge difference in
> > > importance between tests. ClassBuilder and AtomMorph do not have the
> > > same importance. What criteria should apply? how "Core" the tested
> > > functionality is, whether we are talking about enhanced documentation or
> > > newly fubared functionality (the first would obviously not stop a
> > > release..). Any ideas on how to order the test suites to reflect this?
> > > 
> > Should be possible somehow... by building the TestSuits by hand, and
> > having the testserver run those... 
> I thought of something more like having several test packages, where one
> is for version acceptance tests (breakage in which is critical), and
> extra tests (to which people can add stuff that isn't that critical). I
> guess it doesn't matter much whether this knowledge is in test suites or
> in packages.
> 
> > What I was allready planning to do was to set up some conventions for
> > letting the tests print some comment of the status. e.g. lots of
> > the bugs allready have FIXes that just havn't been harvested yet.
> How can you link between a test and it's status?
> 
> Daniel
> 

-- 
Marcus Denker marcus at ira.uka.de  -- Squeak! http://squeak.de