16 Jan
2014
16 Jan
'14
11:15 p.m.
Hi Stefan, I'm reverting the patch in CLN. Please see Bruno's explanation below. (His email hasn't appeared on the list. I have no idea why.) I realized that the patch in CLN is wrong while I was trying to adjust the documentation. Then Bruno's email came in and settled the case. Regarding the original problem in GiNaC: I'm not sure it should be fixed at all. Considering Bruno's argument, it is neither CLN nor GiNaC that is wrong, but the calling code of Li2. It should be more careful when some of the x are computed as +/-1+0.0i. In any case, Li2 shouldn't convert the argument to +/-1 automatically. Please consider rewriting this calling code. Best wishes -richy. -- Richard B. Kreckel <http://www.ginac.de/~kreckel/>