--- On Sun, 3/22/09, Erik Huelsmann <ehuels@...> wrote:
> I forgot to mention: it would be great if we found 1 or 2 common
> topics between these 140 or so failing tests: it's easier
> and more fun to hunt down bugs which are caused by a common
> pattern error instead of 140 single cases.
I suspect the failed integrals are probably mostly the
same bug, maybe another special variable bug ...
There are failed numerical tests as well. In part these
might be due to different conventions for math functions
(e.g. maybe branch cuts) and in part maybe due to faulty
formulas or algorithms. I suspect the numerical problems
are much easier to fix than the conjectured remaining
special variable problems ....
I'll keep you posted if I make any progress.
There 's light at the end of the tunnel!