a bug of "plotode"?

[parisse]

Try with a precise interval in x

Code:

plotode(x*sin(y),[x=-10..10,y],[-2,2])

otherwise autoscaling fails. Indeed the general solution is

Code:

desolve(y'=x*sin(y))

Code:

2*atan(exp(x^2/2)/c_0)+2*n_1*pi

therefore the limit (2k+1)*pi is reached very fast (because of the exponential), and the y-scaling algorithm fails.