@acer does this mean that method=ftoc returns undefined if computing the limit of the antiderivative fails? That's very confusing behavior - I thought undefined meant that maple had concluded the result is actually ill-defined, like dividing infinity by infinity or something.

Also, it's surprising to me that the normal limit function fails, the antiderivative seems like a relatively straightforward function to take the limit of, for example using the asymptotic expansion for polylogarithms here:

https://en.wikipedia.org/wiki/Polylogarithm#Asymptotic_expansions

Anyway, thanks a lot for the tip of using MultiSeries:-limit! That obviously gives a much more useful and convenient result than the polylogarithms I originally came up with.