Question: Why does Maple do so badly at solving ODEs?

Simple question:

Solve y''(x)-y'(x)+2y(x)=10e^(-x)sin(x)

Mathematica answer:

which is correct.

Maple 15 can get the homogeneous solution (the first two terms) but the particular solution? Not as we know it.

y(x) = exp((1/2)*x)*sin((1/2)*sqrt(7)*x)*_C2+exp((1/2)*x)*cos((1/2)*sqrt(7)*x)*_C1+(80/7)*((((1/2-1/2*I-(1/2*I)*ln(e)^2+(1-1/2*I)*ln(e))*sqrt(7)+3/2-1/2*I+ln(e)^3+(3/2+I)*ln(e)^2+(7/2+I)*ln(e))*cos((1/2)*sqrt(7)*x)+sin((1/2)*sqrt(7)*x)*((1/2+1/2*I+(1/2)*ln(e)^2+(1/2+I)*ln(e))*sqrt(7)+1/2+3/2*I+I*ln(e)^3+(-1+3/2*I)*ln(e)^2+(-1+7/2*I)*ln(e)))*exp(-(1/2*I)*x*(-2+sqrt(7)))+(((1/2+1/2*I+(1/2*I)*ln(e)^2+(1+1/2*I)*ln(e))*sqrt(7)-3/2-1/2*I-ln(e)^3+(-3/2+I)*ln(e)^2+(-7/2+I)*ln(e))*cos((1/2)*sqrt(7)*x)-((1/2-1/2*I+(1/2)*ln(e)^2+(1/2-I)*ln(e))*sqrt(7)-1/2+3/2*I+I*ln(e)^3+(1+3/2*I)*ln(e)^2+(1+7/2*I)*ln(e))*sin((1/2)*sqrt(7)*x))*exp(-(1/2*I)*x*(2+sqrt(7)))+(((1/2+1/2*I+(1/2*I)*ln(e)^2+(1+1/2*I)*ln(e))*sqrt(7)+3/2+1/2*I+ln(e)^3+(3/2-I)*ln(e)^2+(7/2-I)*ln(e))*cos((1/2)*sqrt(7)*x)-sin((1/2)*sqrt(7)*x)*((-1/2+1/2*I-(1/2)*ln(e)^2+(-1/2+I)*ln(e))*sqrt(7)-1/2+3/2*I+I*ln(e)^3+(1+3/2*I)*ln(e)^2+(1+7/2*I)*ln(e)))*exp((1/2*I)*x*(-2+sqrt(7)))+exp((1/2*I)*x*(2+sqrt(7)))*(((1/2-1/2*I-(1/2*I)*ln(e)^2+(1-1/2*I)*ln(e))*sqrt(7)-3/2+1/2*I-ln(e)^3+(-3/2-I)*ln(e)^2+(-7/2-I)*ln(e))*cos((1/2)*sqrt(7)*x)+((-1/2-1/2*I-(1/2)*ln(e)^2+(-1/2-I)*ln(e))*sqrt(7)+1/2+3/2*I+I*ln(e)^3+(-1+3/2*I)*ln(e)^2+(-1+7/2*I)*ln(e))*sin((1/2)*sqrt(7)*x)))*sqrt(7)*e^(-x)/((-2*ln(e)+I*sqrt(7)-1+2*I)*(-2*ln(e)+I*sqrt(7)-1-2*I)*(1-2*I+2*ln(e)+I*sqrt(7))*(1+2*I+2*ln(e)+I*sqrt(7)))

Why for example does Maple not know how to evaluate ln(e)? I see ln(e) all of the time in solutions to DEs in Maple.

Why can't Maple do a better job of solving a relatively simple 2nd order DE?