[Freeassociation-devel] SVN fails regression tests?

[crass at berlios.de]

On Mon, 29 Mar 2010 22:45:46 +0200
Arnout Engelen <libical at bzzt.net> wrote:

> On Sun, Mar 28, 2010 at 07:42:16AM -0400, IGnatius T Foobar wrote:
> >  >Anyone noticed that the regression tests fail in SVN?  I think
> >  >we should have a policy that all code commits cause no tests to
> >  >fail (after we get the tests to a point where they are passing).
> >  >Anyone else have opinions on this?  Is there any reason not to
> >  >fix the regression tests?  
> >   
> >  They were broken when we (the current maintainers) originally
> > inherited the code.  While I agree that it needs to be fixed, I'm
> > not quite sure where to begin. 
> 
> As a start, I'd say it would be a good idea to seperate the tests
> that are known to succeed from those that are known to fail. That way
> it is clear failures in the first set are regressions, while the
> second set contains of things that still need looking at.

Yep, I'm thinking we ifdef failing tests so they don't normally compile
and run.  But they can be easily used by defining the right macro
name.  I'll be committing this soon.

> I started looking at the first failure once, too, which indeed was a
> timezone issue:
> 
>  http://sourceforge.net/mailarchive/forum.php?thread_name=1259701956.20184.450.camel%40pohly-mobl1.ikn.intel.com&forum_name=freeassociation-devel
>
> I sort of gave up because it's not quite clear how this should be
> tackled - I didn't consider myself familiar enough with the code to
> make a meaningful proposal (beyond 'have
> icaltimezone_get_builtin_timezone return a timezone with full
> historic data, and provide a helper function to allow clients to
> simplify it if they prefer').

I think the tests which fail due to DST changes should be changed to
"reasonably" get rid of any possible changes to DST.  For instance, I
don't see a change in DST crossing December 15 or July 1 (ie. stay
away from times close to DST.)

Glenn