>>>>> "Richard" == Richard Fateman writes:
Richard> I haven't followed this closely, partly because the code was never
Richard> very clean and tlimit was generally much better. But maybe there
Richard> is another issue worth pointing out, namely that the limit program
Richard> may be oriented to solving real limits, and that it makes mistakes
Richard> in the complex plane not because of bugs, but because it wasn't
Richard> thought about in the original design.
Could be. I'd have to read the thesis again. I also guess that if
that's true, whoever added the code to convert atan to log didn't know
that either.
I also never heard of tlimit. In this particular case
tlimit(atan(x)/(1/exp(1)-exp(-(1+x)^2)),x,inf,plus)
returns 0. Probably because
taylor(atan(x)/(1/exp(1)-exp(-(1+x)^2)),x,inf,1)
returns
0 + ...
Ray