That just looks like an untested edge case. solve! goes to tspan[end], and then step! could continue from there. Or step! could do all of it, but I think if you step! to the end then continue, it might hit a weird floating point bug since it’ll be 1e-16 away from the true end or something like that. Worth an issue.
FWIW, you should probably just be doing tspan = (0.0,Inf) here.