On Thu, 15 Mar 2001, Craig A. Berry wrote:
> At 11:49 AM -0800 3/15/01, Peter Prymmer wrote:
> >So does it look worthwhile to add a special @ivp_tests_to_skip array to
> >test.com?
>
> Perhaps, but the list seems rather too long to justify that
> currently. I wonder if there is any pattern to the remaining
> failures. In particular, I believe 'FAILED at test 0' is often the
> result of failing to find a library. I think I'd rather copy over
> [.lib...] and delete it afterwards than skip that many tests. Thanks
> for working on this.
Yes you've made good points. However, prior to actually conducting the
test I had merely _assumed_ that everything in [.t.lib]*.t would fail. Lo
and behold the [.t.lib]h2ph.t test passed - weird eh? Actually what is
weirder still is that a test of the h2ph utility script is in [.t.lib] and
not in a place like [.t.utils]. I understand that M. Schwern has plans to
rearrange testing directories so perhaps we could take advantage of that.
Back to the topic of an IVP: yes if you could rule out whole classes of
tests, such as e.g. [.t.lib...] then you could avoid putting them into the
saveset thereby decreasing its size as well as making an IVP procedure
easier to maintain I would think.
Peter Prymmer
